Primer of Biostatistics - Term Paper Example

Biostatistics Ahmed Alneyadi Dr. Sabo Bus 226 One Statistics has been described by various people over time in a lot of different but closely linked ways but basically, one can say that statistics is the study of the collection, organization, analysis, interpretation and presentation of data. Statistics is concerned with virtually all dimensions of data, even the planning of data collection in terms of the survey methodologies and experiments therein (Graham, pg.12). Abstract Two Many regard statistics a distinct mathematical science instead of a branch of mathematics. It is a mathematical body that deals with the collection, analysis, interpretation or explanation and presentation of data. During data analysis, quite often, either of descriptive statistics or inferential statistics methodologies is employed. Descriptive statistics (which unlike inferential statistics is not developed on the basis of probability theory) basically refers to the discipline of quantitatively describing the main features of a collection of data while inferential statistics on the other hand denotes the process of arriving at conclusions from data open to random or sampling variations (such as observational errors) (Graham, pg.40). Abstract Three Statistics is a wide field of study, probably one of the broadest disciplines available, with a plethora of applications too. The topic of ‘Fields of Application of Statistics’ is a rich topic in itself and one that sparks a lot of interest both to scholars and professions globally. Some of the sub-topics herein include actuarial science, astro-statistics, business analytics, chemo-metrics, demography, econometrics, environmental statistics, epidemiology, geo statistics, operations research, population ecology, statistical thermodynamics, and biostatistics and so on (Graham, pg.82). This paper deals with the biostatistics aspect of statistics. Introduction Biostatistics, commonly referred to as biometry, is the use of statistics in a wide range of topics in biology. Biostatistics deals with the design of biological experiments (such as in medicine, agriculture, pharmacy, fishery, etc), the collection, summarization and analysis of collected information from the experiments, followed by the interpretation of, and interference from, the results. Biostatistics is a combination of mathematics and reasoning, it considers how research questions are generated, studies are designed, data are collected and results interpreted. If the samples one takes are representative of the population of interest, they will provide good estimates concerning the overall population (Hubert and Carter, pg.90). In biostatistics, one analyzes samples in order to make inferences about the population. In an instance, it would be essential to make comparisons among groups of subjects in order to determine whether certain tendencies such as smoking or even exercise are associated with a greater risk of particular health conditions/ results/ outcomes (Hubert and Carter, pg.156). Biostatistics, like any other discipline has undergone several changes and modifications over time to finally be what it currently is. There are a number of scientists and biologists credited with this development such as Walter Weldon, Karl Pearson, Charles Davenport, William Bateson, Sir Ronald A Fisher, Sewall Wright, Thomas Hunt, and so on (Hubert and Carter, pg.200). In addition, the advent of modern computer technology and affordable computing services has enabled several advances in modern biostatistics which have enabled computer-intensive biostatistics methods such as bootstrapping and re-sampling techniques, alongside new biomedical technologies such as the use microarrays, random forests, gene set enrichment analysis (GSEA), next generation sequencers (for genomics) and mass spectrometry (for proteomics) (Alcala, pg.377). Discussion Biostatistics, among other things, deals with populations and samples. A population is the set of all measurements of interest to a researcher. Populations can either be conceptual or existing. Existing populations are well defined sets of data with elements that are explicitly identifiable. Examples of such existing populations include the CD4 counts of people diagnosed with HIV as of May 2014 in any particular region or the presence/ absence of prior myocardial infractions in males between 40 and 60 years of age in a certain area. Conceptual populations on the other hand refer to non-existing yet visualized or imaginable sets of measurements such as the presence/ absence of myocardial infractions in all current and future high blood patients who receive short acting calcium channel blockers or bio-availabilities of a drug’s oral dose in all healthy subjects under similar circumstances. Samples, preferably collected randomly from populations, describe and make inferences regarding the populations from which they are taken (Alcala, pg.411). There exist two main types of variables in biostatistics – quantitative (numeric) and qualitative (categorical) variables. For instance, CD4 count represents the number of CD4 lymphocytes per liter of blood and is hence numeric (quantitative). Prior myocardial infractions status, for example, can be grouped as categorical if classified as either ‘yes’ or ‘no’, while it can similarly be grouped as numerical if classified as number of prior myocardial infractions. Numeric variables can either be continuous (values falling anywhere corresponding to points on a line segment such as weight and diastolic blood pressure) or discrete (can only take finite/ count-ably infinite number of outcomes such as number of previous myocardial infractions). Similarly, categorical variables can either be nominal (have distinct levels with no inherent ordering such as hair color or sex) or ordinal (do not adhere to a distinct ordering such as the extent of change in patient condition, for example, vast improvement) (Arora and Malhan, pg.312). For instance, In the case of numeric measurements, suppose we have n measurements in our sample, and we label them y1, y2,…yn. Then, we compute the sample mean, variance, standard deviation, and coefficient of variation as follows: ˆμ = y =(∑i=1yi)/n = (y1+y2=…+yn)/n Variance = (∑i=1(yi-y av.)(yi-y av.))/n-1= (square (y1-y av.)+ square (y2-y av.)+ ..square (yn-y av.))/ (n-1) CV = (s/y av.) 100% S = √s squared In addition, there can be dependent and independent variables where dependent variables denote those variables measured as the outcome of interest while independent variables are those variables that determine the population a measurement generates from. Parameters in biostatistics refer to numerical descriptive measures corresponding to populations (Williams, pg.324). These parameters, since populations are not observable, are often considered unknown constants. For numeric variables, there are two commonly reported types of descriptive measures. These are location and dispersion. Measures of location describe the level of the ‘typical’ measurement. Two measures widely studied under the measures of location are the mean (μ) and the median. The mean represents the arithmetic average of all measurements in the population where as the median represents the point where half the measurements fall above it, and half the measurements fall below it (Williams, pg.367). Two measures of the dispersion/ spread, of measurements in a population are the variance and the range. The variance is the average squared distance of the measurements from the mean. Closely related to the variance is the standard deviation. The range is the difference between the largest and smallest measurements. The most common parameter, for categorical variables, is the proportion having the characteristic of interest (when the variable has two levels). Other parameters that make use of population proportions are the relative risk and odds ratios (Williams, pg.444). Under basic probability, in biostatistics, probability is used to measure the chances or likelihood of certain events/ outcomes of an experiment. Under probability, the intersection of events A and B is the event that both A and B occurs whereas the union of events A and B is the event that either A or B occurs. The complement of event A is the event that A does not take place. For diagnostic tests, some of the frequently used probabilities include sensitivity (the probability that an infected individual will correctly test positive based on the diagnostic test, denoted ; P(T+|D+).), specificity (probability that a healthy individual will correctly test negative based on the diagnostic test, denoted by P(T−|D−).), positive predictive value (probability that an individual who has tested positive on a diagnostic test actually has the infection denoted by P(D+|T+).), negative predictive value (the probability that an individual who has tested negative on a diagnostic test actually doesn’t have the disease denoted by P(D−|T−).) and overall accuracy (probability that a randomly selected individual is correctly diagnosed by the test) (Alcala, pg.418). In biostatistics, there exist a number of study designs. In observational studies, the investigator observes subjects, classifies them depending on levels of one or more explanatory variables and a response of interest. Observational studies are further divided into case/ control, cohort and cross-sectional studies. Case–control studies are generally retrospective, and encompass the identification of subjects depending on the level of their response variable, and measuring the level of their explanatory variable (often thought of as some form of exposure) whereas cohort studies, on the other hand are generally prospective involving the identification of subjects depending on the level of their explanatory variable, and obtaining the corresponding response outcome (Lewis, pg.950). Cross-sectional studies involve sampling subjects at random from a population and determining the levels of their explanatory and response variables. Cross-sectional studies are usually conducted retrospectively based on large medical databases at health organizations, states, or at national levels (Arora and Malhan, pg.413). In experimental studies, investigators intervene on their subjects. There are two kinds of experimental studies. These are randomized control trials (RCTs) (which comprise controlled studies where subjects are selected from a population of patients who meet some physical criteria and randomly assigned into treatment groups) and historical control studies (which involve the use of subjects who have been previously treated or not, from which information is used to compare a currently tested treatment) (Arora and Malhan, pg.469). Biostatistics also involves calculations of reliability and validity. Reliability is basically the extent to which results of a test can be replicated if the tests were re-administered to the same person under the same circumstances (it determines how consistent two measures are). Validity can either be internal or external. Internal validity is the degree that we can deduce that changes in the independent variable(s) cause changes in the response variable whereas external validity denotes the extent that the observed experimental results regarding cause and effect can be generalized to other settings or populations (Lewis, pg.809). Hypothesis testing in biostatistics aims at demonstrating that a new drug is better than a placebo control. Hypothesis testing is thus the choice between two propositions concerning an unknown parameter’s value (Lewis, pg.901). Conclusion and Personal Opinion Biostatistics remains one of the most important aspects of statistics as it relates directly to human life and health. The knowledge of biostatistics has since time immemorial assisted mankind to develop and maintain effective treatment and healthcare regimes. Biostatistics and the knowledge amassed there-in in useful in a myriad of medical and other fields (Glantz, pg.345). Among other areas, biostatistics has demonstrated irreplaceable applications in the field of public health, including epidemiology, health services research, nutrition, environmental health and healthcare policy and management. Biostatistics has also been found vital in the design and analysis of clinical trials in medicine, ecology, ecological forecasting, biological sequence analysis and systems biology for gene network inference or pathways analysis. The subject has also played critical role in the analysis of genomics data, for instance from microarray/ proteomics experiments (often concerning diseases or disease levels) as well as in population genetics and statistical genetics in order to link variation in genotype with variation in phenotype. In the latter field of application, the knowledge has been applied specifically in agriculture to enhance crops and farm animals (animal breeding). In biomedical research, knowledge in biostatistics can help in identifying candidates for gene alleles that can cause or influence predisposition to disease in human genetics (Glantz, pg.365). Biostatistics and statistics in general, is therefore a ripe field, open for further exploration for the improvement of lifestyles, health and the general well-being of the larger societies. In addition, there exist a multitude of career opportunities available to students who chose to major in the field of biostatistics (Glantz, pg.400). Works Cited Alcala, Katelynn. Biostatistics. Delhi: Research World, 2012. Print, 377, 411, 418 Arora, P. N., and P. K. Malhan. Biostatistics. Mumbai [India: Himalaya Pub. House, 2010. Print, 312, 413, 469 Glantz, Stanton A.. Primer of biostatistics. 6th ed. New York: McGraw-Hill Medical Pub., 2005. Print, 345, 365, 400 Graham, Alan. Statistics. Blacklick, Ohio: McGraw-Hill, 2003. Print, 12, 40, 82 Hubert, J. J., and E. M. Carter. Biostatistics. Dubuque, Iowa: Kendall :, 1980. Print, 90, 156, 222 Lewis, Alvin Edward. Biostatistics. New York: Reinhold Pub. Corp., 1966. Print, 809, 901, 950 Williams, Brian. Biostatistics. London: Chapman & Hall, 1993. Print, 324, 367, 444 Read More