Pretest Self-Assessment and Review: Preventive Medicine and Public Health. Home · Pretest . DOWNLOAD PDF .. Biostatistics in Clinical Medicine, 3rd ed. In this part of the article, you will be able to access file of Preventive Medicine and Public Health PreTest Self-Assessment and Review 9th Edition PDF. Preventive and social medicine park free ccna todd lammle 7th edition pdf pdf preview windows 7. Preventive and social medicine precalculus mathematics for calculus fifth edition pdf pretest pediatrics twelfth edition pdf. Personality test document headway pre-intermediate 3rd.

Pretest Preventive Medicine 7th Edition Pdf

Language:English, Japanese, French
Genre:Health & Fitness
Published (Last):11.08.2016
ePub File Size:23.64 MB
PDF File Size:19.44 MB
Distribution:Free* [*Registration needed]
Uploaded by: LEONEL

PreTest Emergency Medicine, 3rd Edition () [PDF] 5 MB PDF FREE Physical Diagnosis PreTest Self Assessment and Review 7th Edition [PDF]. February. Preventive Medicine & Public Health: PreTest Self-Assessment and Review: Public Health and Preventive Medicine: Fifteenth Edition (Maxcy-. This is the seventh edition of a self-assessment book in neurology. The first edition was published in This book is primarily directed at medical students .

The investigator concluded that there is an association between diabetes and obesity. The mortality experience of this cohort was no different from that of the general population. The authors concluded that the diagnosis of acute anxiety neurosis is not associated with a decrease in longevity.

The authors concluded that bacterial endocarditis is rare in childhood. The expected utility, life expectancy, is expressed in quality-adjusted life years, or QALYs.

Which of the following statements is true concerning the creation of a decision tree used for clinical decision making? If radiation therapy was never associated with the complication of proctitis, the quality-adjusted life expectancy would be a. The first nodes in a tree are chance nodes b. Branches from the chance nodes must be mutually exclusive and collectively exhaustive c.

The terminal nodes represent prevalence of disease d. The expected utilities are calculated by folding back the tree from left to right e. The numerical values of the expected utility are expressed in different units than the expected outcomes a. The quality-adjusted life expectancy for surgery is a. Based on the results of this decision analysis, which approach appears preferable? Surgery b. Radiation c. Surgery, only if there is no probability of death d.

Radiation followed by surgery, if there is a recurrence e. No preferable approach can be identified 5. Match the name of the parameter below with the appropriate formula. Suicide Served in Vietnam Served elsewhere a. The odds ratio. The relative risk. The excess risk of suicide in Vietnam veterans. The overall incidence per five years of suicide in the study. Page 30 Systematic sampling Paired sampling Simple random sampling Stratified sampling Cluster sampling Each individual of the total group has an equal chance of being selected.

Households are selected at random, and every person in each household is included in the sample. Individuals are initially assembled according to some order in a group and then individuals are selected according to some constant determinant; for instance, every fourth subject is selected. Individuals are divided into subgroups on the basis of specified characteristics and then random samples are selected from each subgroup. Half of the women who are tested have a positive test.

Match the epidemiologic terms below with the correct percentage. Sensitivity of the test. Specificity of the test. Prevalence of chlamydial infection in that community. Predictive value of a positive test. Predictive value of a negative test. Ecologic fallacy Type 1 error Type 2 error Selection bias Misclassification bias The odds ratio for smoking is 3. Dichotomous scale Nominal scale Ordinal scale Interval scale Ratio scale Survival of a particular patient for at least five years.

Birth weight. Frequency of somnolence during biochemistry lectures: Type of medical specialty. Year of birth. Vera Blues, a noted psychiatric epidemiologist, is interested in the diagnosis of depression. She develops a new test for its diagnosis, which she calls the Blues test. Blues applies her new test to persons diagnosed as being depressed by the gold standard; 80 have a positive Blues test. She finds persons who are not depressed; 60 have a positive test. Match the statements that Dr.

Blues made in her article with the appropriate percentage. The probability of a positive DNA-probe. The probability of infection if the DNA-probe is negative. The probability of no infection if the DNA-probe is positive. The answer is d. The multiplicative rule applies to independent events. The probability of two negative consecutive tests is 0. The probability that a woman who has cancer will test negative decreases with the number of mammographies done.

This is inherent to the sensitivity of the test. The higher the sensitivity, the lower the probability of falsenegative tests as they are repeated. The answer is c.

You might also like: API 520 9TH EDITION PDF

This study is a prospective cohort study because the subjects pregnant women were categorized on the basis of exposure or lack of exposure to a risk factor smoking during pregnancy , and then were followed to determine if a particular outcome low-birth-weight babies resulted. The term cohort refers to the group of subjects who are followed forward in time to see which ones develop the outcome. Clinical trials are prospective studies in which an intervention is applied—no intervention was mentioned in the question: In a casecontrol study of the relationship between low birth weight and maternal smoking, infants would be selected on the basis of low birth weight cases and normal birth weight controls and then the frequency of maternal smoking would be compared in the two groups.

In cross-sectional studies, exposure and outcome are measured at the same point in time. A retrospective cohort study is similar in design to a prospective cohort study subjects are chosen on the basis of exposure then assessed for outcome: The answer is b.

A much more likely explanation is regression toward the mean. By referring patients for the study based on a single measurement, those in whom the measurement was high which proved later not to reflect the actual BP are much more likely to be referred than those in whom the measurement was too low.

Neither baseline drift which occurs with measurements on certain machines that require frequent calibration nor measurement error is as likely an explanation.

The Hawthorne effect refers to a tendency among study subjects to change simply because they are being studied. It is much more likely to affect studies of behavior or attitudes than a study of blood pressure.

Person-years of observation are frequently used in the denominator of incidence rates and provide a method of dealing with variable follow-up periods. Person-years of observation simultaneously take into account the number of persons under observation and the duration of observation of each person.

For example, if eight new cases of diabetes occurred among people followed for two years, the incidence would be 8 cases per person-years, or 4 per person-years of follow-up. The distinction between rates and proportions is not well maintained in standard epidemiologic terminology.

Rates should have units of inverse time and will vary depending on the units of measurement of time; they can vary from 0 to infinity. However, such terms as case fatality rate, attack rate, and prevalence rate are in widespread usage even though technically they are all proportions; that is, they vary between 0 and 1 and are unitless. In this case, The answer is e.

The correct values for mean, median, and mode are 3. The mean is the average: The median is the middle observation in a series of ordered observations, that is, the 50th percentile.

In this case, there is an uneven number 9 of observations.

Preventive Medicine and Public Health PreTest Self-Assessment and Review 9th Edition PDF

When the observations are ordered— 1, 2, 2, 2, 3, 4, 4, 6, 7—the median is 3. If the number of observations is even, it is midway between the two middle observations. For example, if we were to have only 8 observations such as: The mode is the observation that occurs with greatest frequency; in this case it is 2, which occurs three times.

In this case, the expected counts are 4. In general, it is used when the sample size is small. According to the table, 10 new cases of tuberculosis developed among the persons belonging to households with a case of tuberculosis at the time of the first survey.

Because these persons were followed for 2 years, the number of personyears of exposure is Therefore, the incidence rate is calculated as follows: Ten new cases of tuberculosis developed among 10, persons belonging to households that had no culture-positive cases at the time of the first survey.

Since these 10, persons were followed for 2 years, the number of person-years of exposure is 20, The relative risk is the ratio of the incidence of a disease in a group exposed to a factor in this case, household contact with tuberculosis to the incidence in a group not exposed to the factor persons without household contact. In experimental studies, the investigators determine exposure of the study and control groups to a suspected causal factor and measure responses in the two groups.

In observational studies, investigators have no control over exposure to a suspected causal factor but can measure responses in those who are and are not exposed.

In both types of studies, the attempt is made to use study and control groups similar in regard to all variables except exposure to the factor under study.

Recall bias, a form of information bias and differential misclassification, occurs when cases are more likely to recall past events than controls. Indeed, persons experiencing a bad outcome may be more likely to search their past and prod their memory about potential causes for the occurrence. This is a particular problem with case-control studies. Recall bias could cause a falsely high odds ratio; it is potentially a problem when using maternal recall to investigate exposures associated with birth defects.

In this case, mothers with children with undescended testes may be more accurate in quantifying smoking habits. Because this misclassification of exposure is not random in both the case and controls, it is termed differential misclassification. Nondifferential misclassification occurs when the memory of an exposure is unrelated to the fact that a person has a disease or not. It is often the consequence of an imprecise measurement of exposure remembering specific nutrition information that occurred many months ago.

The important point to remember is that differential misclassification may result in an overestimate of an association while nondifferential misclassification nearly always causes the results to move toward the null no association.

Selection bias refers to systematic errors in the way subjects are included in a study. Confounding occurs when the apparent effect of an exposure is partly or entirely due to a third factor associated with both exposure and outcome. Although a third factor could potentially be present, it has not been identified here, and the major concern in this case should be the recall bias. Since undescended testes are uncommon, the odds ratio in this study approximates the relative risk risk ratio. Confidence intervals describe the range of values not significantly different from the observed value, with a type 1 error rate alpha of 1.

Thus, if the study is accurate, it suggests that baby boys whose mothers smoke are 2. A larger sample size decreases variability, thus decreasing the confidence interval. For two events or conditions, the probability that either will occur is the sum of their probabilities, minus the probability that both will occur. This is illustrated in the following figure. Ther efore, the probability of A and B must be subtracted from the sum of the probabilities: In this question, it is specifically stated that the two conditions are independent.

When that is the case, the probability that both will occur is the product of their probabilities: The answer to this problem is 0. Note that another common situation is when two conditions are mutually exclusive rather than independent i. In this case, the probability that either one will occur is simply the sum of their probabilities.

For example, if condition A were blue eyes and condition B brown eyes, the probability of either blue or brown eyes would be 0. Case-control studies are well suited to studying rare disorders with multiple potential causes.

The large database will enhance selection of a control group.

Tổng hợp các sách y khoa thế giới All Medical books USMLE

A retrospective cohort study requires that you identify the exposed and the unexposed from years back and that you follow them over time. It is unsuited for rare cancers, and not applicable to the data you have available. Rapidly progressive cancers will be less likely to be detected by a screening test if symptoms rapidly develop because the window period between the time the cancer can be detected when it is asymptomatic by a screening test and the time it will become clinically apparent is short.

This is described as length bias. Screening tests are more effective in terms of prolonging life or other desirable outcomes when they are used to detect more slowly growing tumors. Lead-time bias occurs when the screening test advances the time of diagnosis, but no true prolongation of life occurs because survival for persons who are screened and those who are not is the same from the time the cancer occurs.

Information bias occurs when there is a systematic difference in the way data are collected inaccurate or imprecise measure for either the exposure or the outcome. Recall bias is one form of information bias see answer for question 12 and refers to what one may remember for an exposure, so it is irrelevant here. Selection bias occurs when the inclusion of a subject in a study group is linked to the exposure of interest.

As an example for a case-control study, if women who use oral contraceptives are suspected more often of having deep vein thrombosis DVT , they would be hospitalized more often for evaluation and diagnosed more often than controls. Selection bias can also occur in cohort studies and is related to differential loss to follow-up. Surveillance bias refers to overdetection of the disease of interest because one of the groups goes to the doctor or has a diagnostic test more often than does another group.

For example, women who take postmenopausal estrogens presumably go to the doctor and probably have mammograms more frequently than women who do not; thus, women who take estrogens may be more likely to have breast cancers detected because of the increased surveillance.

Sensitivity and specificity are measures of how often a diagnostic test gives the correct answer. These definitions can be illustrated as follows: Thus, among those with disease, sensitivity measures how often the test gives the right answer. A good way to remember sensitivity is by the initials PID: Similarly, among people who do not have the disease, there are also two possibilities: Thus, specificity measures how often the test gives the right answer among those who do not have the disease.

A good way to remember specificity is by the initials NIH: But predictive value is a little tricky because it also depends on the prevalence of the disease in the population tested. In this case, Dr. Stewells assembled groups of patients with and without cholera, and the prevalence is not given. In this study of patients with profuse diarrhea, of them had cholera.

Note here that this predictive value refers to the predictive value of the test in patients admitted to the hospital with profuse diarrhea.

Since prevalence data from the general population are still lacking, the usefulness of this test in the general population is undefined. Predictably, the positive predictive value of this test in an asymptomatic population will be less. As the prevalence falls, more and more of those tested will not have cholera.

This would change neither the sensitivity nor specificity of the test, which do not depend on disease prevalence, but would affect predictive value: This makes sense: The answer is a. Randomization is the use of a predetermined plan of allocation or assignment of subjects to treatment groups such that assignment occurs solely by chance.

It is used to eliminate bias on the part of the investigator and the subject in the choice of treatment group.


The goal of randomization is to allow chance to distribute unknown sources of biologic variability equally to the treatment and control groups. However, because chance does determine assignment, significant differences between the groups may arise, especially if the number of subjects is small.

Therefore, whenever randomization is used, the comparability of the treatment groups should be assessed to determine whether or not balance was achieved The study described was a case-control study.

In this type of study, people who have a disease cases are compared with people whom they closely resemble except for the presence of the disease under study controls. Cases and controls are then studied for the frequency of exposure to a suspected risk factor.

Matching is a method to control for confounding in case-control studies to eliminate the effect of any extraneous variable that is not under study but may have an effect on the results. Correlation studies are used to compare disease frequencies between entire populations as opposed to individuals.

For example, a correlation study could examine the consumption of animal fat and the rates of colon cancer among 20 different countries. The answers are b, d, a. The relative risk is defined as the incidence rate among the exposed Ie divided by the incidence rate among the nonexposed Io. We can calculate this by using the population attributable risk PAR , which is defined by the attributable risk x prevalence of exposure in the population.

Hennekens, pp — Intent-to-treat analysis, that is including in the final results all the subjects who were initially randomized to receive either the drug or the placebo, is the preferred method of analysis for intervention studies.

Although it may be tempting to include only those who complied with the medication, the results can be misleading. This study is a classic example of this pittfall. Indeed, the study showed that the difference in mortality between those who did and did not adhere to placebo was even greater: The difference persisted even after controlling for 40 different confounders. Thus, something related to compliance with either the medication or the placebo appeared to decrease mortality. Therefore, as a rule, remember that once randomization has been performed, all participants, regardless of their compliance, should be included in the results.

Matching is a technique used in the design of the study to control for confounding. Subjects enrolled in a study are matched for age, gender, smoking, or any variable that is not being analyzed. This technique is not used for large cohort studies as it would often be too time-consuming, restrictive, and expensive to find a match for each subject entering the study.

Therefore, controlling for confounding is done in the analysis when a large group is recruited. Matching is mainly used when dealing with small case-control studies where the number of subjects enrolled would be too small to yield statistical results if stratified by subgroups.

Randomization is used in clinical trials to control confounding sample size needs to be large—see the answer to question Matching cannot be used in correlation studies or cross-sectional studies: The answers are b, a.

In order to maximize compliance, a researcher can assess the compliance of subjects by giving them either the active or inert medication for a certain period of time, before the randomization for the study has occurred. Noncompliant persons can be dropped from the entire study population and the compliant persons are then randominzed to receive either active or inert medication placebo. Keeping logs and frequent contacts from research staff can also help maintain compliance.

The use of the placebo is to assess for responses that may simply be attributed to receiving an intervention, whether active or inert. It has been shown that even patients who receive inert medication can do better than if receiving nothing. This reduces the bias in the ascertainment of outcome. Intent-to-treat analysis refers to including all subjects who were initially randomized in the final analysis of results, compliers and noncompliers alike see the answer to question As this study is likely to require a long follow-up period, every effort must be made to ensure complete follow-up.

The incidence of lung cancer will not affect the internal validity of the study, but if it is low, it may affect the power of the study to measure differences between groups because there may be an insufficient number of outcomes to reach statistical significance between the two groups. Comparison of crude death rates of countries with different population compositions is fruitless.

Adjusting both crude death rates to a standard population gives ageadjusted rates, which can be compared. Developed nations have higher crude death rates because larger proportions of their populations are elderly and thus have a higher probability of dying. Since rates account for population size, a larger population can be compared with a smaller one. Death rates are just one factor in evaluating health care systems.

Hennekens, pp 20— Because the association between the risk factor use of smokeless tobacco and the disease oral lesions is measured at a single point in time in a whole group of subjects, this is a cross-sectional study. A case-control study might be performed over a similar time period, but the sampling would be different: In a cohort study, the habits of a group of players initially free of the disease would be measured, and these players would be followed over time to see how many develop the lesions.

A clinical trial involves allocation of the subjects by the investigator usually randomly to one of two or more treatment groups. Cross-sectional studies allow one to estimate the prevalence the number of existing cases at one point in time divided by the population at risk but not incidence number of new cases occurring over a period of time divided by the population at risk and the period of time at risk.

The odds ratio applies to case-control studies and cross-sectional studies. Hennekens, p The chi-square test can be used for statistical analysis of categorical data no fractions are possible; number of persons are categorized as ill or not ill, dead or alive, etc. The odds ratio can be used as a measure of association. Because the association between the risk factor use of smokeless tobacco and the disease oral lesions is measured at a single point in time in a whole group of subjects, no temporal association between the exposure and the outcome can be assessed.

Furthermore, as this is not a cohort study in which subjects are chosen on the basis of exposure, there should be no expectations that the number of exposed persons would be similar to those who are not exposed. The importance of blinding, while it usually cannot be overemphasized, is not relevant in a study with total mortality as the end point: Power is not relevant in a study that shows a significant effect. In a randomized study, the percentages of patients who actually had myocardial infarctions should be similar in the two groups.

Total mortality is a much more important end point than mortality from coronary heart disease, but long-term follow-up is absolutely essential in determining whether a therapy is useful. Perhaps the new agent simply postpones mortality by a few days or weeks. The secondary attack rate of a disease is the ratio of the number of cases of a specified disease among persons exposed to index cases divided by the total number so exposed.

According to the data, cases of pertussis occurred among fully immunized children who were exposed to a sibling who had the disease. Jekel, , p The relative risk is the ratio of the incidence rates of two groups who differ by some factor—in this instance, immunization status: In decision analysis, utilities refer to the relative values placed on various outcomes that could be experienced by the patients, not the physicians.

For example, perfect health might be assigned a utility of , and death assigned one of 0. What, then, would the utility be for life with moderate back pain? With careful questioning, one finds that most patients place a higher value on life with disability than would be anticipated.

Different techniques can be used to have persons quantify utilities for a given outcome. All the choices listed are methods to control for confounding. Matching and restriction excluding smokers among cases and controls can be achieved in the initial phase of designing the case-control study and before collecting information. Randomization is used for experimental studies. Once the data is collected, control for confounding can be performed in the analysis by stratification or multivariate analysis if there is a need to control for mutiple variables.

We would then stratify the data by smoking status and calculate the odds ratio OR for each stratum smokers and nonsmokers as demonstrated in the following: In this situation, the adjusted OR would also be different than the crude OR. The answers are d, c, e. The percentage of cases of German measles that were asymptomatic, or subclinical, is calculated by dividing the number of asymptomatic persons by the total number of infected persons.

The information was stratified by age to determine if rates were similar.

Medical Sciences: Books/eBooks

The low attack rate in persons 60 and over suggests that this age group had developed immunity to German measles as a result of prior exposure at least 60 years before since there was uniform susceptibility in persons under The answers are a, d, d. There is a trade-off between sensitivity and specificity of a test because there is overlapping of the normal population and the population with disease for most screening tests.

There will be some persons without cancer who will test positive false-positives and some persons with cancer will test negative false-negatives. The trade-off is that it will be less specific: Therefore, some confirmation of the test by another more specific method will be necessary before we can draw any conclusion. Pagano, pp 15— Line graphs are useful for presenting continuous data over time within different populations. In fact, in most cases, the horizontal axis scale in line graphs represents time in year, months, and so forth.

Frequency polygons are used to illustrate frequency distributions for discrete or continuous data. More than one set of data can be superimposed for comparison. The horizontal axis often represents measure of the variable of interest e. Histograms can also be used for this purpose, one set of data at a time.

The horizontal axis should represent the true limits of intervals between data points upper and lower limit and the vertical axis should begin at zero. A bar chart is used to depict the frequency distribution of nominal or ordinal data. Only one set of data is represented for each chart. Pie charts can be used to illustrate relative frequencies of categorical data.

C Pie chart 0. These curves have the same mode, median, and mean measures of location. However, the spread is different and can be assessed by computing the standard deviation measure of dispersion , which will be different for both curves.

Although a large sample size tends to reduce variation and narrow a curve, it is not a summary numerical measure. This curve is skewed to the left or negatively skewed. Such curves have median values that are larger than the arithmetic mean, and the mean also lies to the left of the median. This occurs when more outlying values are smaller than the mean, or points below the median tend to be further away from the median than points above.

A curve is skewed to the right or positively skewed when the opposite occurs, and the mean lies to the right of the median. This curve only has one mode but is not symmetrical nor normally distributed. Large sample sizes increase the precision of a study and decrease the width of the confidence intervals CI.

If the confidence interval includes one when assessing relative risks or odds ratio, it includes the null value. Therefore, the p value will be higher than 0.

The answers are d, b, c. This is an example of receiver-operator curves, or ROC curves. This is plotted against the sensitivity on the vertical axis y. Each curve can be used to determine the optimal cut-off point for the respective test. In general, the point closest to the upper-left corner, where sensitvity is highest and the false-positive rate is lowest, is chosen as the cut-off. The area under the curve is used to calculate the diagnostic accuracy best combined sensitivity and specificity of the test, that is, the probability of correctly identifying disease or no disease based on the result of the test.

In this example, test C has the largest area under the curve compared to the other tests, and therefore would have the greatest diagnostic accuracy.

Sensitivity analysis is used in decision analysis to determine how much impact different probabilities of a particular event will have on the choice of choosing one intervention over another. Computer programs can compute and plot these data. The maximum quality-adjusted life expectancy or years or QALYs for surgery is 4. QALYs are plotted for radiation therapy and surgery for different probabilities of mortality from surgery.

As expected, mortality from surgery does not impact the QALYs obtained from radiation therapy. However, as mortality from surgery increases, the QALYs for that intervention decrease.

If mortality did not impact QALYs for surgery, you would obtain a straight line with the y coordinate at 4. The threshold is the point at which both interventions intersect: The sensitvity analysis from this example demonstrates that mortality rate from surgery is an important variable for determining the best strategy.

The null hypothesis the odds ratio equals one is not rejected. The confidence interval includes 1, and the p value is higher than 0. If the alpha is set at 0. This would therefore narrow the width of the confidence interval. Conversely, if we were to choose an alpha at 0. Colon cancer incidence rate per , The ultimate objective is to predict the value of an outcome based on the fixed value of an explanatory variable.

In this example, we would be able to predict the glomerular filtration rate from a particular value of plasma creatinine, and thus determine what is considered to be within normal limits. Multiple regression is used when we wish to examine the relationship between multiple dependant variables and the independent variable.

Logistic regression is used when y the dependent variable is not a continuous variable, but rather a dichotomous variable for example, presence or absence of disease. Correlation analysis is used to determine whether there is a linear relationship between continuous variables that are treated symmetrically. It would not identify relationships that are nonlinear. It does not imply a cause and effect relationship, nor does it describe the nature of the relationship.

It is used to analyze relationships in correlation studies of population. Is there a linear relationship between fat consumption and colon cancer, immunization rates and infant mortality? Each country represents a point in the plot with a combination of outcomes x,y. Both studies have the same relative risk 0. For study A: These two measures are useful to determine the magnitude of the effect of a given intervention.

These measures can be misleading in assessing the clinical relevance of an intervetion because the overall impact of the intervention is highly dependent on the rate of mortality in the control. In study A, the rate in the control group is very low. Thus, even if the relative risk is very high, the intervention is associated with little overall gain. It is the arithmetic difference between the two groups, or the same as attributable risk. We say reduction if the intervention reduces the risk and increase if the intervention increases a particular risk not meaning a bad outcome.

For study A, the ARR is equal to a reduction of 0. This measure gives a better picture of the impact of an intervention and how much benefit can be attributed to the intervention. We can see that the intervention used in study B would provide more benefit than the intervention used in study A. This gives us an estimate of how many patients will need the intervention before we can avoid one bad outcome, and can be useful for clinicians to get a perspective on the intervention in their practices.

However, the relative risk for men is different than for women. We conclude that gender is an effect modifier. Effect modification is a different concept than confounding. Effect modification provides important information: It is not related to the fact that there may be more men than women in one group or another.

A third factor can be both a confounder and an effect modifier if the adjusted risk differs from the crude, in addition to having different risks in women and in men. It may be neither a confounder nor an effect modifer if the adjusted and crude risks are the same and if the rates in men and women were the same. Finally, it could be only a confounder if the crude and adjusted risks differ, but the rates between men and women are the same.

Stratification can be used to evaluate both confounding and effect modification: The Kappa statistic is often used for reliability studies. For example, it can be used to assess interrater reliability, such as comparing the readings of mammography between different radiologists. It could also be used to assess intrarater reliability, such as comparing responses from participants on surveys given more than once over a period of time to evaluate reproducibility of responses.

The chi-square will not give the degree of association and is used for categorical data. The student t test and correlation studies are used to analyze continuous data. The Wilcoxon rank sum test as well as the Wilcoxon signed rank test and the signed tests can be used when we cannot assume that the underlying population is of normal distribution, especially when dealing with small samples.

The signed test and the signed rank tests are the counterparts for nonparametric distributions of the paired t test, and the rank sum is analoguous to the t test for independent samples.

A drawback of nonparametric methods is that they have less power than the methods used when normal distribution is assumed. The chi-square test is used for categorical data. Analysis of variance is used to test the difference between the means of more than two independent samples. Hennekens and Buring, pp 64— The longer the duration of the disease, the more likely it is to be present at any given time.

If a disease has a high mortality rate short duration , it is unlikely to be counted at any time. Prevalence of disease will increase when a new treatment decreases mortality. A high incidence of disease may or may not have an impact on prevalence: Pagano, p The degrees of freedom for the chisquare distribution are calculated as follows: This is an example of a Kaplan-Meier method, also called the product-limit method, of estimating survival.

Some may be lost to follow-up prior to failure move away, refuse to continue to participte any longer, etc. These are called censored observations incomplete observation of a time to failure. Kaplan-Meier curves appear like uneven steps. Other methods can be used actuarial method , but the Kaplan-Meier is the most frequent. The answers are d, c, b, e. Fetal mortality is defined as the number of stillbirths per births of gestational age greater than 28 weeks.

It evaluates fetal losses of the third trimester. Maternal mortality refers to the death of a woman from any cause related to or aggravated by pregnancy or its management.

Indirect maternal mortality relates to conditions aggravated or caused by pregnancy, labor, or puerperium diabetes, congenital heart disease, etc. The answers are b, c. Greenberg et al. Internal validity can be questioned if there is systematic nonrandom error in the way information is collected. Systematic errors include bias and confounding.

External validity refers to whether the results internally valid of a study can be applied to the other populations. This is a question of judging whether the subjects in the study are similar to the population you are interested in applying the results to such as patients in your clinical practice. Power refers to the capability of a study to detect statistically significant results. Reliability is synonymous with precision: Lack of precision is often due to small sample sizes.

The answers are b, a, d, c. Use of the student t test to assess the difference between the mean systolic pressures of pregnant and nonpregnant women would be appropriate since the two groups are independent samples and the outcome variable is quantitative continuous and approximately normally distributed.

In the study comparing the occurrence of hepatitis B surface antigen in medical and dental students, use of chi-square analysis would be appropriate because both the predictor and outcome variables are categorical and dichotomous; that is, students are classified by the presence or absence of the antigen and by medical or dental student status.

The McNemar test is used for a matched pair of categorical data. Analysis of variance will permit evaluation of the effects and interaction of sex and drug on the glucose level. Use of the regular student two-sample t test in this instance is inappropriate because the two samples are not independent— the same subjects are in each.

The answers are b, a, e, f. Effect modification occurs when one factor modifies the effect on outcome of another.

As an example, a high bilirubin seems to be a much stronger risk factor for bilirubin-induced brain damage if the baby is sick in other ways see question Confounding occurs when the association between two variables is distorted by the fact that both are associated with a third. For example, the association between coffee and lung cancer is distorted by smoking: Similarly, lead levels need to be related to IQ separately at each level of socioeconomic status to assure that the association is not due to confounding.

The possibility that hyperactive children have high lead levels because they are hyperactive, rather than vice versa, is not confounding; it is simply a case in which the direction of causality is turned around effect-cause. Nondifferential misclassification results in the mixing of two groups because the measure of either the exposure or the outcome was imprecise, for example, assessing precise diet habits by questionnaires in a case-control study, and going back many years.

Most people are unlikely to remember what and how much they ate years ago, and thus exposures will be similar in the cases and controls. Recall bias, a form of differential misclassification, is unlikely in this setting. Nondifferential misclassification always biases results toward the null value. Lead-time bias refers to a distortion of the apparent efficacy of a screening program see answer to question The answers are a, c, b, e.

The point prevalence is the proportion of people in a population who have a disease at a given point in time. The numerator is the number of existing cases of a disease; the denominator is the total population at risk of the disease at that point in time. The standardized mortality or morbidity ratio SMR is the ratio of the observed number to the expected number of deaths or cases of the disease.

For example, age-specific rates of angina pectoris in nonsmokers can be applied to the age distribution of smokers to obtain the expected number of cases of angina pectoris in the smokers. The SMR of smokers for angina pectoris is the observed number of cases divided by the expected number so calculated.

The cumulative incidence is the number of new cases of a disease that occur in a period of time divided by the population at risk during that time. The incidence density takes into consideration the length of time subjects participated in the study and the denominator is expressed in person-time of observation. The relative risk or risk ratio is the incidence of disease in subjects with a risk factor divided by the incidence in those in whom the factor is absent.

The denominator is not the incidence in the general population because then subjects with the risk factor would be included. If the risk factor is uncommon and the relative risk is close to 1. However, other risk factors, for instance, a relative with CHD, are quite prevalent. The term relative risk can be confusing when the risk factor has to do with being a relative of a patient; in this instance, risk ratio is a preferable synonym.

The answers are c, e, d, b, a. Although these terms are usually applied to epidemiological studies, they are also applicable to examples from everyday life. Lead-time bias commonly refers to the apparent increase in life expectancy seen in patients who have their disease diagnosed with a screening test. The same would be true of a study that found that anatomy students lived at the same address for a longer period of time than fourth-year medical students, most of whom move to start internships.

The study would not be wrong, but any conclusions that suggested that anatomy students are more stable than fourth-year clerks would be meaningless. Recall bias classically refers to a situation in which persons with a disease are more likely to remember an exposure say, to a toxic chemical than persons who are healthy. This is part of a human tendency to look for explanations for bad outcomes—like failing an examination. A type 1 error occurs when a result is found to be statistically significant by chance in a sample even though there is no effect in the population.

Power is the chance of finding an effect in your sample if it truly exists in the population. One problem with finding out that your friends have been out at the movies is that they may not tell you the truth recall bias , or you may ask the wrong ones, such as those sitting next to you in the library surveillance bias. So you can give yourself credit if you made one of those choices as well, assuming you understood what you were doing! The answers are b, e, a, c.

Greenberg, pp 22—23, 49— The case fatality rate is a measure of the severity of the disease. It is a ratio of the number of deaths caused by a disease to the total number of cases of that disease and is usually expressed as a percentage. The crude mortality equals the total number of deaths from all causes during a year divided by the average population at risk during that year. It is usually expressed as the number of deaths per people. The secondary attack rate is a measure of the contagiousness of an infectious disease.

The numerator is the number of cases of disease in contacts of the index case; the denominator is the number of contacts exposed to the index case during a specified period. Rates of disease are called morbidity rates. The answers are b, a, e, b. Matching is a way of selecting subjects that are comparable with respect to specific variables. For example, in a casecontrol study, a control could be selected that is the same age and sex as the case.

It is thus a sampling strategy to achieve comparability among groups. Stratification is an analysis strategy with the same purpose. For example, survival could be compared separately in different age strata, as in question This might be important if the subjects with high renin levels were also older than the subjects with low levels, since a difference in survival between the two groups might be due to age, rather than to differing renin levels.

Age adjustment takes stratification by age one step further. After mortality or another parameter is calculated for specific age strata, it is combined in a weighted average to yield a single number.

The weights used are the sizes of the different age strata in a standard population. Age adjustment is used more often for comparing mortality in populations with differing age structures. Multivariate statistical analysis, like stratification, is an analysis technique for achieving comparability among groups. It involves modeling the associations between variables in order to allow their different effects to be isolated from each other.

For example, in multiple regression, the relationships between variables are modeled as a straight line. Survival analysis is a technique by which persons followed for variable lengths of time are counted according to the length of time they were followed. For example, in the cohort study of renin levels mentioned previously, instead of simply comparing the proportions surviving five years, the cumulative probability of survival could be plotted for the two groups, and the two curves compared.

The Kaplan-Meier and life table analysis are two methods used for survival analysis. The first plots the percentage of persons alive after each year since a diagnosis. The answers are d, a, b, d. Hennekens, pp 58, , For proper comparison of the frequency of a disease in two groups, the rate of disease, not the number of cases, must be compared. The number of cases may reflect the age structure of the population served by the hospital. Age-specific attack rates that incorporate the number of cases in each age group, divided by the number of persons in each group, should be calculated.

In order to determine that an association between two conditions such as diabetes and obesity exists, an investigator must show that obesity is significantly more common in persons who have diabetes than in persons who do not have diabetes. The controls are necessary in order to test the significance of the association and must resemble the cases as closely as possible in all ways except for the absence of the disease under study.

Because death may be a major reason for loss to follow-up, the most conservative approach is to assume that everyone lost to follow-up has died. The conclusion in question 98 is invalid because of the lack of denominators to calculate the rate of bacterial endocarditis in different age groups.

In addition, the autopsy series merely gives an estimate of the proportion of deaths in different age groups, not the frequency of occurrence of endocarditis with age. The autopsy series may also be invalid as a source of data from which to draw conclusions because of factors that determined whether an autopsy was performed.

The answers are b, c, a, e. When constructing a decision analysis tree, the first node is a decision node to reflect the choices you have to make to manage a specific medical problem.

Branches from the chance nodes must reflect all possibilities. Terminal nodes reflect the outcomes or utilities assigned to the outcomes, such as death, survival, quality-adjusted life years, and so forth.

Thus, the utilities are expressed in the same units as the outcomes e. Therefore, surgery provides more QALYs than radiation therapy. The answers are a, d, e, c. Hennekens, pp 73— When the disease is rare, the odds ratio closely approximates the relative risk.

However, the study in the example is a cohort study, so relative risk can be calculated directly from the table. Excess risk is defined as the difference between the risk in those with the risk factor and those in whom it is absent. Whereas the relative risk and odds ratio are unitless since any measurements of time in the denominators cancel out , the excess risk must have an explicit or implied time period in the denominator.

Thus, if the yearly risk of suicide was 0. Note that a more precise way to measure the incidence, relative risk, and so on would be to use person-years at risk in the denominators, but this leads to greater computational and conceptual complexity. The answers are c, e, a, d. Pagano, pp — Simple random sampling is a process in which individuals are sampled independently, and each individual of the population has an equal probability of being selected. In cluster sampling, groups of people e.

A common analytic mistake is to pretend that subjects obtained in a cluster sample were obtained in a simple random sample. This can lead to incorrect results because the subjects are not sampled independently. Systematic sampling is a process that first requires the arrangement of the group to be sampled in some kind of order. Then individuals are selected systematically throughout the series on the basis of a predetermined sampling fraction or constant determinant, for example, every fifth, tenth, or hundredth person in the ordered group.

Although systematic sampling may seem almost the same as simple random sampling, it is much less desirable. In stratified sampling, a population is divided into subgroups based on defined characteristics such as age, sex, or severity of illness, or any combination of these; then random samples are selected from each subgroup.

For example, you could take a random sample from a group of to yearolds, from a group of to year-olds and from a group of to 29year-olds from a total population of to year-olds. This is used particularly in situations where the distribution of each subgroup is not uniform in the group as a whole for instance, there may be only a few to year-olds, and they may be missed if you were to use a simple random sample of the to year-old group.

This method allows you to make sure that persons from each subgroup are represented in your sample. In paired sampling, or matching, selection of one or more controls for each case is based on age, sex, time, time sequence, geographic location, or some other defined relationship to the case so it is not random.

For example, selection could be based on the next patient admitted after each case, the sibling nearest in age to each case, or the person who lives closest geographically to each case. The answers are d, c, a, b, e. Since we were not told how many women were tested, we can just make up a number—say, We are told that half have a positive test: Test positive Test negative Infection No Infection? Sensitivity is simply the proportion of women with chlamydia who will have a positive test remember: Specificity is the proportion of women without chlamydia who will have a negative test remember: The answers are c, a, b, d.

Any time a study fails to achieve statistical significance, a crucial question to ask is whether the study had enough subjects. Although subjects per group followed for five years seems like a large number, only a tiny minority perhaps 10 per group would be expected to have a myocardial infarction. Thus, the sample size in this instance may have been inadequate to detect a meaningful difference between the groups. The ecologic fallacy occurs when associations among groups of subjects are mistakenly assumed to hold for individuals.

Thus, although among communities, high rates of condom use may be associated with higher fertility rates perhaps because condom use acts as a marker for sexual activity in general , among those who use the condoms, the fertility rate could in fact be zero. A type 1 error occurs when, just by chance, a statistically significant difference between groups is found.

Studies attempting to correlate multiple risk factors with multiple diseases particularly when there is no good biologic reason to suspect an association are especially prone to type 1 errors. Looking for associations separately in different subgroups compounds the problem. Selection bias occurs when the subjects selected for the study are somehow not representative of the population from which they come.

Thus, since patients with lung cancer will be mostly smokers, smokers will be overrepresented among the controls, and smoking will look like a weaker risk factor than it really is. The answers are a, c, e, b, d. Pagano, pp 7— The scale of measurement is an important determinant of the amount of information in a variable and the type of statistical analysis that can be used. Dichotomous variables like sex have only two possible values. Some variables may be artificially dichotomized, with subsequent loss of information.

For example, a patient either survives five years or not; thus survival to five years is an example of a dichotomous variable. The variable could be made more informative, however, if the actual number of months of survival was specified. Nominal variables have more than two possible values, but no intrinsic ordering. Nominal and ordinal are often confused. Thus, one value cannot really be subtracted from another.

An example is dates of birth: Ratio scales are measurements in relation to a clear zero point. Thus, measurements on ratio scales can be meaningfully divided by each other. For example, one baby may weigh twice as much as another or have twice as high a platelet count.

Absolute temperature is measured on a ratio scale, whereas temperature in Fahrenheit or Celsius is measured on an interval scale. The answers are a, b, e.

Answering the first two of these questions is easiest if the results of Dr. This is the same as the likelihood that a person with depression will have a positive Blues test. The answers are c, i, j. Elsevier; 3 edition January 28, Language: Retaining the underlying principles of the original editions this comprehensive rewrite and re-presentation provides complete coverage of orthopaedic trauma surgery as relevant to contemporary practice.

Baron Aslan. Job-Ferreira Jobbiomedico. Vallejo, Steven A.

Abrams https: George Karamalis. Pre-Obstetric Emergency Training: A Practical Approach. Thanks a lot. Research Methodology for Health Professionals Open University Press; 2nd Revised ed.

A beginner's guide to evidence-based practice in health and social care Thank you guys!!! Can you check please the file Something wrong. Snap Wechat. Andrey , I downloaded both files already and it's ok to unzip. Just re-download Thank you!!!!!.He has not previously received therapy for tuberculosis in the past nor has he had any known contact with persons infected with tuberculosis.

Although it may be tempting to include only those who complied with the medication, the results can be misleading. The parents heard her scream, ran up to her room, and shooed the bat out the window. Antibiotic prophylaxis of adults and children Items — A year-old man presents with a single, indurated, painless ulcer on the penis that appeared two days ago. Sensitivity is simply the proportion of women with chlamydia who will have a positive test remember: