Home 
Contact 
Profile 
Videos 
Synopsis for use in teaching sessions of the postgraduate course ‘Essentials of Epidemiology in Public Health’ Department of Social and Preventive Medicine, Faculty of Medicine, University Malaya Malaysia July 20th 2009
MODULE OUTLINE
6.1 MEASURES OF ASSOCIATION and EFFECT
6.1.1 General Concepts
6.1.2 Tests of Association
6.1.5 Meta Analysis
6.2 SOURCES AND TREATMENT OF BIAS
6.2.5 Survey Error and Sampling Bias
6.3 HEALTH STATUS INFORMATION
6.3.3 Disease Registries with Cancer as an Example
6.4.1 Health Economics
6.5 READING AND WRITING SCIENTIFIC LITERATURE
6.5.1 Literature Search
6.5.2 Critical Reading of a Journal Article
6.5.3 Abuse or Misuse of Statistics
6.5.4 Scientific Writing
6.5.5 Scientific Publishing
UNIT 6.1
MEASURES OF ASSOCIATION and EFFECT
Data analysis affects practical decisions. It involves construction of hypotheses and testing them. The 2-sided test covers both p1 > p2 and p2 > p1. The 1-sided test covers either p1 > p2 or p2 > p1 and not both. The 2-sided test is preferentially used because it is more conservative. Simple manual inspection of the data is needed can help identify outliers, assess the normality of data, identify commonsense relationships, and alert the investigator to errors in computer analysis. Data models for continuous data can be straight line regression, non-linear regression, or trends. Data models for categorical data are the maximum likelihood and the logistic models. Two procedures are employed in analytic epidemiology: test for association and measures of effect. The test for association is done first. The assessment of the effect measures is done after finding an association. Effect measures are useless in situations in which tests for association are negative. The common tests for association are: t-test, F test, chi-square, the linear correlation coefficient, and the linear regression coefficient. The effect measures commonly employed are: Odds Ratio, Risk Ratio, Rate difference. Measures of trend can discover relationships that are too small to be picked up by association and effect measures.
6.1.2 TESTS OF ASSOCIATION
The tests of association for continuous data are the t-test, the F-test, the correlation coefficient, and the regression coefficient. The t-test is used for two sample means. Analysis of variance, ANOVA (F test) is used for more than 2 sample means. 1-way ANOVA involves one factor (explanatory variable). 2-way ANOVA involves 2 factors. Multiple analysis of variance, MANOVA, is used to test for more than 2 factors. Linear regression is used in conjunction with the t test for data that requires modeling. Dummy variables in the regression model can be used to control for confounding factors like age and sex.
The common test of association for discrete data is the chi square test. The chisquare test is used to test association of 2 or more proportions in contingency tables. The exact test is used to test proportions for small sample sizes. The Mantel-Haenszel chi-square statistic is used to test for association in stratified 2 x 2 tables. The chi square statistic is valid in one of the following conditions: (a) if at least 80% of cells have more than 5 observed, (b) if at least 80% of cells have more than 1.0 expected, (c) if there are at least 5 observed in 80% of cells. If the observations are not independent of one another as in paired or matched studies, the McNemar chisquare test is used instead of the usual Pearson chisquare test. The chisquare works best for approximately Gaussian distributions.
An epidemiological study should be considered as a sort of measurement with parameters for validity, precision, and reliability. Validity is a measure of accuracy. Precision measures variation in the estimate. Reliability is reproducibility. Bias is defined technically as the situation in which the expectation of the parameter is not zero. Bias may move the effect parameter away from the null value or toward the null value. In negative bias the parameter estimate is below the true parameter. In positive bias the parameter estimate is above the true parameter. A study is not valid if it is biased. Systematic errors lead to bias and therefore invalid parameter estimates. Random errors lead to imprecise parameter estimates. Internal validity is concerned with the results of each individual study. Internal validity is impaired by study bias. External validity is generalizability of results. Traditionally results are generalized if the sample is representative of the population. In practice generalizability is achieved by looking at results of several studies each of which is individually internally valid. It is therefore not the objective of each individual study to be generalizable because that would require assembling a representative sample. Precision is a measure for lack of random error. An effect measure with a narrow confidence interval is said to be precise. An effect measure with a wide confidence interval in imprecise. Precision is increased in three ways: increasing the study size, increasing study efficiency, and care taken in measurement of variables to decrease mistakes.
Meta analysis refers to methods used to combine data from more than one study to produce a quantitative summary statistic. Meta analysis enables computation of an effect estimate for a larger number of study subjects thus enabling picking up statistical significance that would be missed if analysis were based on small individual studies. Meta analysis also enables study of variation across several population subgroups since it involves several individual studies carried out in various countries and populations. Criteria must be set for what articles to include or exclude. Information is abstracted from the articles on a standardized data abstract form with standard outcome, exposure, confounder, or effect modifying variables. The first step is to display the effect measures with each article with their 95% confidence limits to get a general idea of their distribution before proceeding to compute summary measures. The summary effect measure, OR or b, is computed from the effect measures of individual studies using weighted logistic regression or computing a MH weighted average in which the weight of each measure is the inverse of its precision i.e. 1/(se)2. In both the logistic or MH procedures, each study is treated as a stratum. The combined effect measure is then statistically adjusted for confounding, selection, and misclassification biases. Tests of homogeneity can be carried out before computing the summary effect measure. Sensitivity analysis is undertaken to test the robustness of the combined effect measure.
UNIT 6.2
SOURCES and TREATMENT OF BIAS
6.2.5 SURVEY ERROR and SAMPLING BIAS
Total survey error is the sum of the sampling error and three non-sampling errors (measurement error, non-response error, and coverage error). Sampling errors are easier to estimate than non-sampling errors. Sampling error decreases with increasing sample size. Non-sampling errors may be systematic like non-coverage of the whole sample or they may be non-systematic. Non-systematic errors cause severe bias. Sampling bias, positive or negative, arises when results from the sample are consistently wrong (biased) away from the true population parameter. The sources of bias are: incomplete or inappropriate sampling frame, use of a wrong sampling unit, non-response bias, measurement bias, coverage bias, and sampling bias.
UNIT 6.3
HEALTH STATUS INFORMATION
6.3.1 HOSPITAL INFORMATION SYSTEMS
Medical records departments collect, store, and access information used for health care, financial or administrative reasons. Quality control is needed to eliminate inaccuracy and inconsistencies. Privacy and confidentiality are balanced against the need for timely information by caregivers. SOAPIE, the acronym for problem-oriented medical recording, stands for Subjective complaints, Objective complaints, Assessment, Plan, Intervention, and Evaluation. The electronic medical record (EMR) enables clinical data net-working, direct on-line data entry from terminals in wards, record linkage, and record integration. Hospital records are analyzed for process and outcome performance indicators, planning and projections, cost analysis, assessing access and affordability, and surveillance.
Health care information is used for decision making, problem solving, and planning. It is sourced from demographic data, morbidity data, morbidity data, health care utilization, hospital care records, outpatient treatment records, environmental monitoring, occupational monitoring, health care activities, needs assessment, disease registers, health surveys, injury and accident monitoring, and vital statistics.
6.3.3 DISEASE REGISTRIES WITH CANCER AS AN EXAMPLE
Cancer registration is continuing, systematic collection of data on reportable neoplasms. The data is comprehensive including socio-demographic, clinical, laboratory, radiological, and treatment variables. The hospital cancer registry helps physician in follow-up of patients, sends data to outside registries (community, regional, or national cancer registries), and is used for hospital-based epidemiological studies (incidence and prevalence, immediate causes of death, survival, and treatment). The hospital cancer registry is a good source of controls for case control studies of cancer etiology. Disease registries are also available for other diseases.
Vital statistics are from mandatory reporting. They generate hypotheses for further investigation. They are analyzed in conjunction with ecologic or environmental data. The following statistics can be computed from vital statistics.
Crude Birth Rate (CBR) is births per 100,000 of mid-year population per year. Premature Birth rate is births at gestation age 28-38 weeks per 1000 live births per year. Low birth weight rate is births < 2500 g per 1000 live births per year. Very low birth weight rate is births < 1500 g per 1000 live births per year.
Crude Death Rate (CDR) is deaths in a year per 100,000 of mid-year population. Proportional Mortality Ratio (PMR) is deaths of a specified kind as a proportion of the total number of deaths. Case-fatality ratio is the proportion of deaths from persons with a specified disease condition. Fetal death rate is deaths >= 20 weeks per 1000 births (live births + still births) per year. Fetal death ratio is death =<20 weeks per 1000 1000 live births. The abortion ratio is number of induced abortions per 1000 live births. The Infant Mortality Rate (IMR), the most important indicator of community health, is deaths at ages 0-12 months per 1000 live births per year. The Total Neonatal Mortality Rate is deaths within 1-28 days of birth per 1000 live births. The Early Neonatal Mortality Rate is deaths within 7 d of birth per 1000 live births per year. The Late Neonatal Mortality Rate is deaths at age 7-28 days of birth per 1000 live births per year. Post Neonatal Mortality Rate is deaths aged 28days -1 year of birth per 1000 live births per year. Peri-natal Mortality Rate is deaths aged > 28 weeks up to 7 days of birth per 1000 total births (live births + stillbirths) per year. The peri-natal mortality ratio is the number of fetal deaths >= 28 weeks + deaths within one week of birth per 1000 live births. Maternal Mortality Rate is deaths in pregnancy or within 42 days of delivery per 100,000 births.
Marriage rate is marriages in a year per 1000 of population. The divorce rate is divorces in a year per 1000 or population or per 1000 marriages.
Total Fertility Rate (TFR) is births per year 1000 women aged 15-44 in mid-year population. Gross reproductive rate is reproductive rate computed for girls only. The net reproductive rate is the proportion of girls surviving to the reproductive age out of 1000 live births. The replacement level is TFR of 2.1.
Population is described by age, sex, race/ethnicity, marital status, education, and occupation. Population pyramids reflect both birth and death rates. They display population structure by age, and gender. Population structure and future projection are determined by fertility, mortality, and migration. Population pyramids have a narrow base and a wide top in industrial countries and a wide base and a narrow top in non-industrialized ones. Atypical pyramids are due to war, genocide, and migration. Demographic shift/transition is change in population structure with falling birth and death rates. Life tables are current or cohort (generational); general or specific (for gender, race, and place); complete or for given ages; and full or abridged (by age groups). They are used for actuarial, pension, & annuities computations; assessing health services; determining life expectancy and survival; and computing Potential Years of Life Lost (PYLL). Life expectancy at birth, a sensitive indicator of the life of the community, indicates current death rates. It is lower at birth than at age 1 year due to high IMR in the first year of life.
UNIT 6.4
6.4.1 HEALTH ECONOMICS
Health economics, an integration of medicine and economics, is application of micro-economic tools to health. Economic concepts used in health economics are scarcity, production, efficiency, effectiveness, efficacy, utility, need, want, demand & supply, elasticity, input-output, competition, marginal values (marginal costs and marginal benefits), diagnosis related group (DRG), service capacity, utilization, equity, value of money (present and future), and compounding & Discounting. Economic analysis uses models and hypothesis testing. The assumptions of a free/competitive market is not always true in health care because of supplier-induced demand and government regulation. Measurements of production costs, health outcomes, the value of human life, and the quality of life are still controversial. The purpose of economic analysis is to evaluate projects regarding cost minimization, cost benefit analysis, cost effectiveness analysis, and cost utility. It also plays a role in decision analysis.
Planning is a circular process that includes: situation analysis, prioritization, goal and objective definition, and choice of strategies, and evaluation. Objectives of health planning can be universal, national, or local. Rational planning is based on analysis of data, defining objectives, and formulating plans to achieve those objectives. In incremental planning plans evolve as problems arise and solutions are found for them. Mixed scanning is a judicious mixture of rational and incremental planning. The methodology of planning is defined by answering the questions about planning: how techniques), who (planning by specialists of the community or both), when (long term or short term), and where (place and whether centralized or decentralized). Planning can be for manpower, facilities, or services. Planning for a new program proceeds by identifying a problem, defining the problem, understanding the problem, planning an intervention, and evaluating the intervention. Needs assessment and prioritization are necessary preliminary steps in planning. Indicators of need are morbidity, mortality, and social deprivation. Needs assessment proceeds in 5 steps: determining present health status, assessing the environment, identifying and prioritizing existing programs, assessing service deficits in light of existing programs, dealing with the problems, and validating the needs. Prioritization considers: extent and seriousness of the problem, availability of effective cure or prevention, appropriateness and efficiency of the cure or prevention, and whether intervention will be at the level of the individual or the level of the community. The goals of public health intervention include: raising awareness of the health problem, increasing knowledge and health skills, changing of attitudes and behavior, increasing access to health care, reducing of risk, and finally improved health status. The following are the public health interventions: behavioral modification, environmental control, legislation, social engineering, biological measures, and screening for early detection and treatment of disease.
The health care system can be described as resources, organization, and management. It is described according to availability, adequacy, accessibility, acceptability, appropriateness, assessibility, accountability, completeness, comprehensiveness, and continuity. It consists of institutions, human resources, information systems, finance, management, and organization, environmental support, and service delivery. Its nature is determined by demographic, cultural, political, social, and economic factors. Health care services include: preventive care, primary care, secondary care, tertiary care, restorative care, and continuing health care (for the elderly). Health care delivery systems can be classified by ownership (profit or not for profit and government or private), method of funding (public taxation, direct payment, or insurance), type of care (western, alternative, or traditional), and level of care (primary, secondary, or tertiary). Modes of health care delivery can be the physician office, a Health Maintenance Organizations (HMO), Preferred Provider Organizations (PPO), or ambulatory Care Centers. Health care personnel are classified as independent providers (physicians), limited care providers (eg dentists), nurses, allied health professionals, and public health professionals. Health care facilities are physician offices, hospitals, nursing homes, out-patient services (ambulatory services), emergency room services, Health care maintenance organizations (HMO), rehabilitation centers, and continuing care facilities.
Quality assurance (QA) is formal and systematic identification, monitoring, and overcoming problems in health care delivery. Quality indicators are mortality, morbidity, patient satisfaction, and various rates. Consensus guidelines, Good Clinical Practice guidelines, clinical protocols, and nursing guidelines are a bench-mark against which clinical performance can be evaluated. QA review may be concurrent or retrospective. The QA reviewers may be independent clinical auditors from outside or may be part of the health care team. QA in hospitals centers around review of the patient charts. The aim of QA review is to ascertain compliance with the given guidelines. If a deviation is found, it is documented as well as its surrounding circumstances. It is discussed at the departmental QA committee. The committee will suggest actions to be taken to alleviate the deficiency and map out an implementation plan. The QA review process is cyclical.
UNIT 6.5
READING and WRITING SCIENTIFIC LITERATURE
6.5.1 LITERATURE SEARCH
A document is stored data in any form: paper, book, letter, message, image, e-mail, voice, and sound. Documents of medical importance are usually journal articles, books, technical reports, or theses. The sources of on-line documents are Medline, on-line journals, on-line books, on-line technical reports, on-line theses and dissertations. MEDLINE was established in 1971. Every year 400,000 articles from 3,700 journals are added and are indexed using medical subject headings (MESH). GRATEFUL MED is a query language used to search MEDLINE. Retrieval technology for formatted character documents is now quite sophisticated. It uses queries that are short documents used to retrieve larger documents by matching, mapping, or use of Boolean logic (AND, OR, NOT). In matching, the most common form of retrieval, the query is matched to the document being sought after determining what terms or expressions are important or significant. The search can be limited by subject matter, language, type of publication, and year of publication.
6.5.2 CRITICAL READING OF A JOURNAL ARTICLE
For critical reading of scientific literature, the reader must be equipped with tools to be able to analyze their methodology and data analysis critically before accepting their conclusions. Common problems in published studies are incomplete documentation, design deficiencies, improper significance testing and interpretation.
The main problem of the title is irrelevance to the body of the article. Problems of the abstract are failure to show the focus of the study and to provide sufficient information to assess the study (design, analysis, and conclusions). Problems of the introduction are failures of the following: stating the reason for the study, reviewing previous studies, indicating potential contribution of the present study, giving the background and historical perspective, stating the study population, and stating the study hypothesis.
Problems of study design are the following: going on a fishing expedition without a prior hypothesis, study design not appropriate for the hypothesis tested, lack of a comparison group, use of an inappropriate comparison group, the Berkson's fallacy, selection of cases and controls from different populations, and sample size not big enough to answer the research questions. The following terms are often confused with one another. ‘Measurement’ is using instruments. ‘Calculation’ deals with numbers and formulas. ‘Estimation’ is used in two senses as an approximation in measurements or as computation of statistical parameters. ‘Determination’ is a general term for getting to a conclusion by use of the 4 methods above. The term ‘study’ is generic and can be confused with experiment that refers to only some types of studies.
Problems in data collection are: missing data due to incomplete coverage, loss of information due to censoring and loss to follow-up, poor documentation of data collection, and methods of data collection inappropriate to the study design.
Problems of data analysis are failures in the following: stating type of hypothesis testing (p value or confidence interval), use of the wrong statistical tests, drawing inappropriate conclusions, use of parametric tests for non-normal data, multiple comparisons or multiple significance testing, assessment of errors, assessment of normality of data, using appropriate scales and tests, using the wrong statistical formula, and confusing continuous and discrete scales.
Problems in reporting results are: selective reporting of favorable results, numerators without denominator, inappropriate denominators, numbers that do not add up, tables not labeled properly or completely, numerical inconsistency (rounding, decimals, and units), stating results as mean +/- 2SD for non-normal data, stating p values as inequalities instead of the exact values, missing degrees of freedom and confidence limits.
Problems of the conclusion are failures in the following: repeating the results section, discussion of the consistency of conclusions with the data and the hypothesis, extrapolations beyond the data, discussing short-comings and limitations of the study, evaluation of statistical conclusions in view of testing errors, assessment of bias (misclassification, selection, and confounding), assessment of precision (lack of random error), and assessment of validity (lack of systematic error).
Internal validity is achieved when the study is internally consistent and the results and conclusions reflect the data. External validity is generalizability (i.e. how far can the findings of the present study be applicable to other situations) and is achieved by several independent studies showing the same result. Inability to detect the outcome of interest due to insufficient period of follow-up, inadequate sample size, and inadequate power.
6.5.3 ABUSE or MISUSE OF STATISTICS
Statistics can be abused by incomplete and inaccurate documentation of results as well as selection of a favorable rate and ignoring unfavorable ones. This is done by 'playing' either with the numerator or the denominator. The scales of numerators and denominators can be made artificially wider or narrower giving false and misleading impressions. Statistical results are misleading in the following situations: (a) violating the principle of parsimony, (b) study objective unclear and not reflected in the study hypothesis (c) fuzzy, inconsistently, and subjective definitions (of cases, non-cases, the exposed, the non-exposed, comparison groups, exposure, method of measurement), (f) incomplete information on response rates and missing data.
6.5.4 SCIENTIFIC WRITING
The goal of scientific writing is clarity. The following must be observed about sentences: short concise sentences, use of personal pronouns, subject-verb agreement is a common mistake, using active and avoiding passive sentences, proper organization of parallel ideas, and proper use of parentheses.
A paragraph must start with a short and simple topic sentence that is an overview of the message contained in that paragraph. Each paragraph should convey only one message. The sentences following the topic sentence provide details and support for the topic sentence. Ideas in a paragraph should be presented in the right order with no missing steps using one of the following alternatives: least to most important, most to least important, concise to the detailed, time chronological order, problem followed by solution, or solution followed by the problem. Links and transitions such as ‘which is’ should be used when moving from one group of ideas to another to ensure continuity in the paragraph. There must be consistency in the order in which information is mentioned. If certain objects were mentioned in a certain order in the introduction, they must be mentioned in the same order all through the writing. The writer should maintain a consistent viewpoint all through the paper and not appear to be jumping from point to point. Important messages must be given emphasis.
The purpose of the title is to identify the main topic or message of the paper so as to attract readers. A good title is unambiguous, concise, and contains important words. It should contain the following: independent variable(s), dependent variable(s), the study subjects or materials, and statement of the main message like ‘to study the effect of’, ‘to determine’ etc.
The abstract is an overview of the report with a few significant details. It should be written to be read by both those who read the full paper and those who do not read the full paper. Normally the abstract should not exceed 250 words. The abstract should mirror the sections of the paper: introduction, materials & methods, results, and discussion. The present tense is used to state the research hypothesis and the answer. The past tense is used for the experiment. An abstract is accompanied by keywords that are used for indexing.
The introduction should be short. It should start with stating the research question or research hypothesis and then go on to elaborate. The transition should be from the known to the unknown and from the big picture to the detail. The introduction should mention the type of study, the study subjects or materials (substances, animals, and persons). In some cases the introduction may briefly mention the proposed experimental approach to answering the research question. Results should not be mentioned in the introduction. The introduction should state whether the work is new or original.
The aim of the materials and methods section is to describe the experimental techniques in detail sufficient for another trained scientist to replicate the procedures. The order of presentation is different for animal and clinical studies. For animal studies the order is: materials and animals, preparation, study design, interventions, methods of measurement, calculations, and data analysis. For clinical studies the order is: study subjects, inclusion criteria, exclusion criteria, study design, interventions, method of measurement, calculations, and data analysis. Independent and dependent variables should be identified. Intermediate results can be put in the materials and methods section. Final results should be put only in the results section. Details of sample size determination should be provided.
The results section presents the findings of the procedures carried out in the methods section. It should be brief and to the point. A distinction must be made between results and data. Result refers to summary information obtained from data analysis. Results of hypothesis-based studies should be in the past tense. Data of descriptive studies should be in the present tense. Data is the actual numerical information often presented in a summarized form. The result is presented followed by presentation of supporting data. Data are presented in the form of tables and diagrams (figures, bar diagrams, graphs, pie-charts, maps etc). Presentation of numerical data in text should be kept to a minimum. Only results relevant to the research hypothesis should be presented. Both negative and positive results are presented. It is considered scientific fraud to present only those results that the author thinks favor a particular hypothesis. The results section is written in chronological order. The most important results are presented before the least important. Magnitude of change should be presented as a summary statistic such as percentage change instead of presenting the raw data. Summary statistics normally used as the mean, the median, and the the proportion. The mean should be presented properly as mean +/- standard deviation or standard error of the mean (SD or SE) with units of measure indicated. Measures of effect are normally the chisquare and the t statistics. Actual p values should be given instead of indicating <0.05 or >0.05. When specifying the sample size the type of sample should be explained for example ‘the sample was 20 rats’ instead of the sample size was 20’. Emphasis can be put on some results and not others. Not all the data from the study need be reported. Citing data in the text takes less space but is more difficult to read. A topic sentence is used to give an overview. Important results are put first.
Figures used to present results must have a strong visual impact and must be simple. The following types of figures are used: line graph, scatter-gram, bar graph, histogram, and the frequency polygon. The title of the figure should reflect its contents. It must be labeled correctly. Symbols must be defined. The names of variables and units of measurement must be labeled appropriately. Tables must be properly titled and column headings clearly indicated. Footnotes, subscripts, and superscripts can be used.
The discussion section states the research hypothesis, answers it, and supports the answers using data from the current study and other studies. It provides reasons to show that the answer to the question is reasonable. It explores and explains possible sources of error and bias. It also identifies and explains differences between the study results and published results. As part of intellectual honesty it discusses the strengths as well as the weaknesses of the study and how they impact on the interpretation of the results. Issues of validity and precision are also addressed. Also discussed is whether the result is new and how important it is.
References are used to acknowledge information obtained from others. The references must be the most recent and most easily available on the subject. Review articles are better than original articles. They may be journal articles, books, PhD theses, abstracts of meetings, or conference proceedings. The reference should be put immediately after the relevant text. If there are several references in a sentence, cite each reference at the relevant point and do not wait to put all of them at the end of the sentence. References should be written using the Vancouver style which is: Author. Title. Journal Year; Volume (number): starting page – ending page.