how to read a scientific paper

The next step, having identified the study objectives and overall design, is to conduct a detailed appraisal of the methods and results. The following six guidelines, each in the form of a question about the research and including a checklist of criteria, are summarised in the box

(1) STUDY DESIGN APPROPRIATE TO OBJECTIVES? (2) STUDY SAMPLE REPRESENTATIVE?

(6) DISTORTING INFLUENCES?

-Extraneous treatments -Contamination -Changes over time -Confounding factors -Distortion reduced by analysis

2

-Is there an alternative interpretation of the data that the author did not address? • How are the findings unique/new/unusual or supportive of other work in the field? • How do these results relate to the work I'm interested in? To other work I've read about? • What are some of the specific applications of the ideas presented here? What are some further experiments that would answer remaining questions?

Reading is an active task. Before and during your reading, ask yourself these questions:

-Who are these authors? What journal is this? Might I question the credibility of the work? • Have I taken the time to understand all the terminology? • Have I gone back to read an article or review that would help me understand this work better? • Am I spending too much time reading the less important parts of this article? • Is there someone I can talk to about confusing parts of this article?

(5) COMPLETENESS?

-compiance -drop outs and deaths -missing data

Introductions serve two purposes:

-creating readers' interest in the subject and -providing them with enough information to understand the article

why read scientific papers

-current (its the most up to date stuff) -replicable (I can redo the experiment myself) -has raw data (save time-use their results) -shows logic (do you beleive it or not)

CONTROL GROUP ACCEPTABLE

-definition of controls -source of controls -matching and randomisation -comparable charectersics -

how to read a scientific paper

-dont read straight through

3. introduction

-explains motivation and importance of research. provides background info -question: do you understnad background info? do you need to look up references for more info?

4. results

-provide raw data you might need for your own research. figures and tables provide the data in a compact format for easy viewing. -question: do you understand what the axes means? what units are used? does the curve make sense?

how to read a scientific paper steps

-read abstract -skim introduction -skip methods (for now) -read results -read intro -read discussion -read methods

starting the appraisal

-review a scrututred abstract -what are the objectives of the research -what is the overall study design -

before you read you need the right equipment

-scientific dictionary (look up terms you dotn know, accessscience.com) -your hand notebook (make notes so you remember insight) -friends/colleagues (explain to others will help you understand the paper yourself)

steps to reading scientific research

-should begin by skimming the article to identify its structure and features. -As you read, look for the author's main points - generate questions before, during, and after reading. -Draw inferences based on your own experiences and knowledge. -And to really improve understanding and recall, take notes as you read.

study deisign appropritate to objectives

-source sample -sampling method -sample size -Entry criteria and exclusions -non responde

2. discussion

-summerize important results -give resons for conclusions based on results -question: do you agree with the logic of the conclusion? are these results useful to you?

(4) QUALITY OF MEASUREMENTS AND OUTCOMES?

-validity -reproductibility -blindness -quality control

Because articles contain so much information, it may be difficult to distinguish the main points of an article from the subordinate points. Fortunately, there are many indicators of the author's main points

-visuals (especially figure and table titles) • first sentence or the last 1-2 sentences of the Introduction -Title • Abstract • Keywords

Critical appraisal of published research: introductory guidelines

....

how to read scientific paper article

.................

Sample size

A statement in the methods section that a sample size was chosen in order to have sufficient power to detect a medically meaningful result at a certain level of statistical significance would normally be adequate evidence that steps had been taken to ensure an appropriate sample size. In the absence of such a statement it may be necessary to seek help from a statistician or an appropriate text 10, 11 to establish whether the sample size was adequate.

REVIEW A STRUCTURED ABSTRACT

An abstract should summarise the research succinctly and include the objectives, study design, important results, and the authors' conclusions. When starting to appraise an article reviewing a structured abstract is particularly helpful because the different components of the research can readily be identified.3, 4 Serious flaws may be detected at this stage.

Contamination

Another problem in controlled trials is contamination, in which one group is affected by another. For example, in a dietary intervention study people in a control group may change their diet because they hear about supposed benefits from dietary changes in the intervention group.

Changes over time

Be wary of studies in which data on a characteristic have been collected from two groups of subjects at different times. Observed differences between the groups might be due to changes in the characteristic or its measurement over time, and not to real differences between the groups.

WHAT IS THE OVERALL STUDY DESIGN?

Before dissecting the methods in detail the overall design of the study should be clarified, as this helps to determine those aspects of the appraisal on which to concentrate. The design should be stated in the abstract and at the beginning of the methods section. Most studies comprise one of six designs: case report, case series, cross sectional, cohort, case-control, and controlled trial.

(1) STUDY DESIGN APPROPRIATE TO OBJECTIVES?

Deciding if the overall study design is appropriate may require more common sense than a detailed knowledge of epidemiological methods. If, for example, the purpose of a study is to evaluate a new treatment a controlled trial is almost imperative, as a trial without a control group would be fraught with difficulties in knowing whether improvement in patients was due to the treatment. Similarly, a project examining prognosis would normally require follow up by means of a cohort study. On the other hand, research investigating the cause of disease might adopt any of the designs shown in the figure.

Distortion reduced by analysis

Distorting influences may also be minimised by some form of stratification or adjustment procedure in the analysis

Confounding factors

Distorting influences may exist in studies examining the association between a risk factor and disease where the purpose is to find out whether the association is real or spurious (caused by a confounding factor influencing both the risk factor and the disease). In such studies it is necessary to account for possible confounding factors. This may be satisfied by matching in the selection of controls or by evidence of comparability between cases and controls.

Blindness

During data collection a common source of bias is that the subjects or those collecting the data are not blind to the purpose of the research. The problems that may occur in controlled trials are well known: subjects, observers, and researchers, by wishing the intervention to succeed, produce an unrealistically good success rate

5. Take notes as you read

Effective readers take notes—it improves recall and comprehension. You may think you'll remember everything you read in researching class assignments, professional papers, proposals, or your thesis, but details will slip away.

(2) STUDY SAMPLE REPRESENTATIVE?

If research is to be applicable and relevant to other populations the study sample (group selected to participate) must be representative of the group from which it is drawn (study population), which in turn should be typical of the wider population to whom the research might apply (target population).

source of sample

If research is to be applicable and relevant to other populations the study sample (group selected to participate) must be representative of the group from which it is drawn (study population), which in turn should be typical of the wider population to whom the research might apply (target population). Appropriateness of the target and study populations is usually a subjective assessment based on our knowledge of the topic under investigation

Source of controls

In case-control and cohort studies the source of controls should be such that the distributions of characteristics (not under direct investigation) are similar to those in the cases or study cohort. For example, in a study of exposure to lead and mental ability in children the source of controls should ideally be a group whose social class distribution is similar to that from which the cases were derived.

Matching and randomisation

In case-control studies cases and controls are often matched for certain characteristics, such as age and sex. Did the matching process seem to have been carried out correctly? In controlled trials, on the other hand, subjects are often randomly allocated to intervention and control groups. The method of randomisation should be assessed to ensure that the subjects were truly randomised - for example, by use of computer generated random numbers

Comparable characteristics

In controlled trials random allocation to intervention and control groups usually leads to comparability, but not necessarily so, and the distributions of age, sex, and other prognostic variables should therefore be compared between the two groups. Similarly, in case-control and cohort studies matching or other methods of selecting controls are not infallible and the comparability of the groups must be assessed.

Non-respondents

In most studies some subjects do not respond to invitations, some refuse to participate, and others do not attend for examination. The response rate is often viewed as an indicator of the representativeness of participants, but the size of response is only one aspect of sampling and may be less important than the comparability between participants and nonrespondents.

Sampling method

In population based studies random sampling is the ideal method of avoiding selection bias and producing a sample typical of the study population. In other studies non-random sampling may be adequate;

Definition of controls

In studies using a comparison or control group it is important to assess whether this group was adequate for the purpose under study. In a case-control study, for example, were the criteria for defining controls appropriate and was the control group checked to ensure that it did not contain cases?

Reproducibility

In the interests of expediency many research projects pay too little attention to the reproducibility of the measurements. Would the same results have been obtained if the measurements had been taken by a different observer or on a different day

mising data

Incomplete results may often occur due to difficulties in obtaining specimens, laboratory tests going awry, and lost data. The extent and nature of the loss must be assessed in order to estimate possible bias. Also, selectivity in reporting of results and the exclusion of data from tables may have an effect on the conclusions that can be drawn from the research. It is worth checking that in addressing the objectives of the study the authors have presented data on the most appropriate measurements and that some have not mysteriously disappeared.

Validity

It is important to assess the validity of measurements made in a research study - that is, the extent to which they reflect the true situation. Dietary questionnaires, for example, are notoriously inaccurate in obtaining a true picture of a person's regular nutritional intake. When a single test is used as a proxy measure of disease the validity of the test (sensitivity and specificity) should be stated in the article. In a randomised controlled trial the results may depend on the measurement of one outcome and it is thus essential that this is an important end point which is sensitive to change

Skim the article and identify its structure.

Most journals use a conventional IMRD structure: An abstract followed by Introduction, Methods, Results, and Discussion. Each of these sections normally contains easily recognized conventional features

4. Draw inferences.

Not everything that you learn from an article is stated explicitly. As you read, rely on your prior knowledge and world experience, as well as the background provided in the article, to draw inferences from the material.

Making a judgment

Once a detailed appraisal of the methods and results has been conducted a decision must be made on whether the methods were adequate and the results clear cut enough so that the objectives were achieved and useful information produced. Unfortunately, there is no magical formula which will convert assessments of detail into an overall score on the worth of a paper. The pros and cons of the research have to be weighed implicitly and a judgment made

quality control

Overall, the extent to which the researchers have instituted quality control measures for the examination of subjects, collection of data, and laboratory tests should give some idea of the likely quality of data. Measures might include testing the accuracy and repeatability of observers, checking the calibration and accuracy of instruments, and random checks for errors in data recording

Entry criteria and exclusions

The criteria for entering subjects into a study must be examined carefully; the stage of disease or time of onset, for example, may have a profound effect on the results of treatment or in the detection of aetiological factors. Exclusion criteria should also be defined appropriately. Furthermore, any description of the study participants must be scrutinised in order to assess whether the sample was representative

compiance

The end results of a study may be incomplete in relation to the number of subjects who were first enrolled. This need not necessarily lead to bias in the results, but careful assessment is required. In controlled trials continuing compliance of subjects with a regimen may be a serious problem and, although this may partly be overcome by carrying out an "intention to treat" analysis

WHAT ARE THE OBJECTIVES OF THE RESEARCH?

The next step is to identify the precise objectives of the research. These should be stated in the introduction to the paper or may be given in the abstract. A principal goal of appraisal is to establish whether research objectives have been met, and the most precise form of the objectives should be sought.

Extraneous treatments

The results of studies are often distorted by outside influences. In controlled trials, for example, a common problem is that subjects may be exposed to treatments in addition to the one being evaluated. Thus in assessing a trial the question has to be asked, "Could there possibly be extraneous treatments which might have influenced the results? Have these been identified in the study and the results interpreted accordingly?"

why read scientific papers 4 you can evaluate the conclusions

articles provide the authors explanation of their results and conclusions. you can see their assumption and determine wheather you belive them or not.

The Discussion also functions to provide a \

clear answer to the question posed in the Introduction and to explain how the results support that conclusion.

The Methods section tells the reader what

experiments were done to answer the question stated in the Introduction. Methods are often difficult to read, especially for graduate students, because of technical language and a level of detail sufficient for another trained scientist to repeat the experiments. However, you can more fully understand the design of the experiments and evaluate their validity by reading the Methods section carefully.

why read scientific papers 3 it has acutual data

if you need to know exact results or properties for your own research... articles include actual data, uncertainties, conditions of the experiment, etc.

The purpose of appraising a paper is to discover if the

methods and results of the research are sufficiently valid to produce useful information. The prime objective is not necessarily to assess the authors

Drop outs and deaths

n cohort studies as well as in controlled trials drop outs and deaths in the study sample may occur. It is important to assess not only the proportion of drop outs in each group but also why they dropped out, as this may give a clue to possible bias. For example, more healthy people may move and be lost to follow up, so that a cohort study excluding them might produce an unrealistically gloomy outcome

why read scientific papers 2 it can be replicated

popular articles and books give you general info and results. scholarly journals give you enough info that you could do the experiment yourself. you can verify the research to see if you get the same results.

The Results section contains

results—statements of what was found, and reference to the data shown in visuals (figures and tables).

Generally, introductions accomplish this by leading readers from broad information (what is known about the topic) to more

specific information (what is not known) to a focal point (what question the authors asked and answered). Thus, authors describe previous work that led to current understanding of the topic (the broad) and then situate their work (the specific) within the field.

1.first stop abstract

tell you briefly what experiment was done and what was found question: what specific results are mentioned? are they relevant?

why read scientific papers 1 its current

textbooks can be years out of date by the time they are published. journals tell you what is happening right now.

The worst way to approach this task is to treat it like

the reading of a textbook—reading from title to literature cited, digesting every word along the way without any reflection or criticism.

Anatomy of a Scientific Paper

title, abstract, introduction, methods, results, discussion, references

After reading, ask yourself these questions:

• What specific problem does this research address? Why is it important? • Is the method used a good one? The best one? • What are the specific findings? Am I able to summarize them in one or two sentences? • Are the findings supported by persuasive evidence?

Abstracts usually contain four kinds of information:

• purpose or rationale of study (why they did it) • methodology (how they did it) • results (what they found) • conclusion (what it means) (You should probably begin reading a paper by reading the abstract carefully and noting the four kinds of information outlined above)