extracting and synthesizing information from a literature review

4
EDITORIAL Extracting and synthesizing information from a literature review Search terms Authorship, literature review, research report, writing. Acknowledgements I wish to acknowledge the PhD students who most recently trialed this extraction and synthesis method for their coursework: Armi Earlam, Lucy Graham, Kristina Hall, and September Nelson. Excerpts from course assignments are used with permission of the students. My students are always my best teachers. doi: 10.1111/jspn.12021 Although there are a number of published models for conducting a literature review, I have never found an explicit method for extracting relevant information from the retrieved literature. Thus, several years ago I began teaching the techniques I developed over the years for my own writing. I always preface this instruction with the caveat that it is one way to approach this task—not the way—and encourage others to modify this method to best fit their own cognitive processing style. THE PROBLEM The problem becomes apparent when one has com- pleted the exhaustive literature review and achieved an intimidating stack of articles to read, usually in a limited time. Often, we find ourselves reading and highlighting the articles, restacking them, and sub- sequently searching again and again for that one piece of information our mind’s eye remembers seeing on the top right-hand side of a page about halfway through an article that now seems to have vanished. In course papers describing their evolving extraction methods, recent doctoral students articu- lated the problem in these words: I have always enjoyed looking for articles, finding ones that answer some pertinent part of my question, print- ing them out, and making The Big Stack. The Big Stack is ridiculously satisfying visual proof of all the work just done to answer a question. Much like clean laundry, it is there neat and tidy and ready to be put away. And much like the way I do laundry, that is where I lose my order and focus, and the neat stacks sit ready to be organized . . . for a very long time. Pile Paralysis. When I finally have the courage to dive in to The Big Stack, I am clumsy, slow, and superbly inefficient. Even in a paper done earlier in this semester, I was, more often than I like to admit, unable to locate relevant information from articles I had just read. (L. Graham, excerpt from course assignment, November 27, 2012) During my master’s thesis research, most of my literature review utilized print versions of articles. I read each, underlined what I found interesting, and then organized them into piles on my living room floor. At various points in research and writing, I re-read pertinent articles and subsequently re-created piles based upon specific ques- tions I was attempting to answer at a given time. This methodology created frustrating article hunts on mul- tiple occasions, as the papers changed piles on a daily basis. As I entered the PhD program, my literature review was still rather haphazard, but had moved to computer- based archiving. Still, articles piled up on my desktop with very little organization or evaluation to provide ready synthesis of findings. As an example, a literature review of allostatic load and cardiovascular disease left me reading or skimming articles a minimum of four times prior to completing the short paper. (K. L. Hall, excerpt from course assignment, November 30, 2012) Thus, the problem is not only one of reading and organizing the literature, but of extracting the spe- cific answers to the search questions, corralling them in an easily retrievable format, and accomplishing the task in a timely manner. As is frequently the Journal for Specialists in Pediatric Nursing 85 Journal for Specialists in Pediatric Nursing 18 (2013) 85–88 © 2013, Wiley Periodicals, Inc.

Upload: roxie-l

Post on 27-Mar-2017

212 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Extracting and synthesizing information from a literature review

E D I T O R I A L

Extracting and synthesizing information from a literature review

Search termsAuthorship, literature review, research report,

writing.

AcknowledgementsI wish to acknowledge the PhD students who

most recently trialed this extraction and

synthesis method for their coursework: Armi

Earlam, Lucy Graham, Kristina Hall, and

September Nelson. Excerpts from course

assignments are used with permission of the

students. My students are always my best

teachers.

doi: 10.1111/jspn.12021

Although there are a number of published modelsfor conducting a literature review, I have neverfound an explicit method for extracting relevantinformation from the retrieved literature. Thus,several years ago I began teaching the techniques Ideveloped over the years for my own writing. Ialways preface this instruction with the caveat that itis one way to approach this task—not the way—andencourage others to modify this method to best fittheir own cognitive processing style.

THE PROBLEM

The problem becomes apparent when one has com-pleted the exhaustive literature review and achievedan intimidating stack of articles to read, usually in alimited time. Often, we find ourselves reading andhighlighting the articles, restacking them, and sub-sequently searching again and again for that onepiece of information our mind’s eye remembersseeing on the top right-hand side of a page abouthalfway through an article that now seems to havevanished. In course papers describing their evolvingextraction methods, recent doctoral students articu-lated the problem in these words:

I have always enjoyed looking for articles, finding onesthat answer some pertinent part of my question, print-ing them out, and making The Big Stack. The Big Stackis ridiculously satisfying visual proof of all the work justdone to answer a question. Much like clean laundry, it isthere neat and tidy and ready to be put away. And much

like the way I do laundry, that is where I lose my orderand focus, and the neat stacks sit ready to be organized. . . for a very long time. Pile Paralysis. When I finallyhave the courage to dive in to The Big Stack, I amclumsy, slow, and superbly inefficient. Even in a paperdone earlier in this semester, I was, more often than Ilike to admit, unable to locate relevant informationfrom articles I had just read. (L. Graham, excerpt fromcourse assignment, November 27, 2012)

During my master’s thesis research, most of my literaturereview utilized print versions of articles. I read each,underlined what I found interesting, and then organizedthem into piles on my living room floor. At various pointsin research and writing, I re-read pertinent articles andsubsequently re-created piles based upon specific ques-tions I was attempting to answer at a given time. Thismethodology created frustrating article hunts on mul-tiple occasions, as the papers changed piles on a dailybasis.As I entered thePhDprogram,my literature reviewwas still rather haphazard, but had moved to computer-based archiving. Still, articles piled up on my desktopwith very little organization or evaluation to provideready synthesis of findings. As an example, a literaturereview of allostatic load and cardiovascular disease leftmereadingor skimmingarticlesaminimumof four timesprior to completing the short paper. (K. L. Hall, excerptfrom course assignment, November 30, 2012)

Thus, the problem is not only one of reading andorganizing the literature, but of extracting the spe-cific answers to the search questions, corralling themin an easily retrievable format, and accomplishingthe task in a timely manner. As is frequently the

bs_bs_banner

Journal for Specialists in Pediatric Nursing

85Journal for Specialists in Pediatric Nursing 18 (2013) 85–88© 2013, Wiley Periodicals, Inc.

Page 2: Extracting and synthesizing information from a literature review

case, clarifying the problem and committing to thesearch for a better process is 90% of the solution.

THE SOLUTION

The solution I will offer here is simple but not alwayseasy. It is designed to achieve the time managementgoal of handling each article only once (or at least asfew times as possible). Essential to this process arewell-designed search or research questions to focusthe extraction of answers.

Step one: find the answers

Do not bother to organize the articles in the stack; ittruly does not matter where you start. Just pick upthe first article in the stack, read it, highlight it, writenotes in the margins or whatever helps you call outthe answers to the search/research questions.Remember that the goal is to glean the importantinformation on the first read. Keep the questions infront of you and stick to those questions andanswers. Do not get sidetracked with other interest-ing information. Mark that other information in adifferent highlighting color or with a P for “parkinglot” or an S for serendipity, but do not let your mindwander from the task at hand. In that sense thisprocess is a bit like meditation, where you allowextraneous thoughts to flow through your mind butyou decline to focus on them. Similarly, do not getsidetracked trying to read every word in the docu-ment. Read as though you were reading the yellowpages. Let your eyes flit past irrelevant headings.Read the first sentence (thesis sentence) of a para-graph and decide if you need to read further. Skim!You will identify more relevant information whenyour mind stays active and focused than if youattempt to painstakingly analyze each article. If youare reading an article from a peer-reviewed journal,rest assured that it has already been analyzed or itwould not be published. Could you find additionallimitations in the article if you thoroughly critiquedit? No doubt. But you might not make it past the firstfew articles in the stack that way.

If it sounds like heresy to recommend skimmingthe articles, let me provide a rationale usingBenner’s (1984) classic novice-to-expert theory.Knowing how to read and critique a scholarly articleis foundational to successful literature review and isa skill usually introduced at the undergraduate level.Initially a novice relies on rules to perform a task, butwith repeated experience the discrete steps and bitsof information integrate to form a more meaningful

whole. With considerable experience, the novicebecomes an expert who can “zero in on the accurateregion of the problem without wasteful consider-ation of a large range of unfruitful” material(Benner, 1984, p. 32). The skill to efficiently skim ascholarly article to glean the “fruit” will necessarilydepend on the reader’s progress in the novice-to-expert continuum. Thus, graduate students wouldbe expected to be much nearer the expert stage thannew baccalaureate graduates. If you still feel like anovice or have only beginning competency inreading scholarly articles, I recommend that youfind colleagues who share your passion for this par-ticular literature search and divide the articlesamong the group to ensure the search is completedbefore the passion fades.

Step two: number the article and recordthe reference

As previously identified, easy retrieval of previouslyread articles is one aspect of the problem. Thus, I rec-ommend numbering each article in the stack andrelentlessly keeping the numbered articles in order.Again, it does not matter which article is numberone; the point is that each article has a number. Thatnumber is then linked to the article’s reference. Ifyou use an electronic reference manager, promi-nently add the article number in the notes section.If you create the reference list manually, numberthe list. You can later easily remove the numbers,remove references you will not use in the paper orreport, and place the remaining references in properorder. This technique assures that each bit ofextracted information is linked to its source and thatyou can easily retrieve that source from the num-bered article stack. It also prevents a frustratingsearch for the reference you neglected to capturewhen it was in your hands.

Step three: extraction

Extraction consists of recording the information youhighlighted within the article. I find that I can onlykeep a couple of articles in my mind at once so Iswitch frequently between reading and recording.Successful extraction requires that the informationbe placed in a useful and retrievable format forwhich I recommend evidence tables and “buckets.”

An evidence table is a systematic list of specificitems of interest. The table itself is usually in land-scape orientation to provide for several columns. Forexample, Neu, Corwin, Lareau, and Marcheggiani-

Editorial R. L. Foster

86 Journal for Specialists in Pediatric Nursing 18 (2013) 85–88© 2013, Wiley Periodicals, Inc.

Page 3: Extracting and synthesizing information from a literature review

Howard (2012) conducted a systematic review tofind the benefits, efficacy, and safety of pharmaco-logic and nonpharmacologic treatment for infantswith gastrointestinal reflux disorder. Evidence tablesdetailing their findings included columns for (a)authors, design, and setting, (b) infant age, gender,and race/ethnicity, (c) inclusion and exclusion crite-ria for the study being reviewed, (d) trial length, (e)intervention and control, (f) outcome statistics, and(g) a narrative column for outcomes. Importantly,beyond noting the authors in the first column, thereis no convention for the number of columns or forthe information in the columns. The column labelsare author-defined to capture information needed toanswer the search questions. As illustrated in theNeu and colleagues (2012) article, the columnscontain material you would expect to find in most orall of the articles being reviewed. Evidence tables areprevalent within the current literature, and a quicksearch for review articles will provide manyexamples.

Evidence tables are one of the reasons I describedthis extraction method as simple but not necessarilyeasy. Evidence tables are somewhat time consum-ing, especially at first. With practice, the reader/author can quickly fill in the columns, however, andthe column labels help direct the elements toretrieve from the article. The PhD student whodescribed her original process as resulting in pileparalysis commented that after trying this methodshe “was not only NOT paralyzed, I was surprisinglymotivated to discover the important information Iwas after, filling in abbreviated facts, essential infor-mation, and necessary data elements into the evi-dence table as I went along” (L. Graham, excerptfrom course assignment, November 27, 2012).

Evidence tables have one major limitation andthat is the size of the columns. Brief information fitsbest. Thus, readers/authors are left wondering whatto do with the sentence they wish to quote or thenotes about the article that will not fit in the table.That is where buckets come in. Buckets are justlabeled categories for corralling larger quantities oftext. They allow the reader to extract anything andeverything else that is relevant, thus providing onemore safeguard against having to go back and readthe article again. Buckets require labels, which arelikely to be somewhat broader in scope than thecolumns in the evidence table. For example, if Neuand colleagues (2012) had employed buckets inextracting information, the labels may have beensomething like Benefits, Efficacy, Safety, Reflux,Study Limitations, and Other Insights. They might

also have reserved a bucket for “parking lot” or “ser-endipitous” material that they wanted to examine ata later date.

Buckets can hold anything you would like toretain for consideration in writing the formal paperor report. They differ from evidence tables not just inthe space available, but in the fact that one does notpublish buckets, whereas one frequently includesevidence tables with an article or formal report.Thus, it is okay for buckets to be less formal andperhaps even a bit messy as long as they clearlyrecord the literature excerpts. They are an excellentplace to record something you wish to quote or areference table copied from an article. The caveat isthat anything placed in a bucket needs to link tothe article number and include quotation marksand page numbers for copied material. Discipliningoneself to be thorough in this process prevents acci-dental plagiarism and eliminates the need to searchfor the page number for a text citation or reference.

Practically speaking, I do not use evidence tablesfor everything I write, especially if I am not referenc-ing that many sources. For a short or less compli-cated manuscript (such as an editorial), I usually usebuckets to capture the reference material I will needas I write. Evidence tables are important to organizeinformation for quick comparison across sourcesand to accompany formal papers and manuscripts.Whereas buckets are companions for complex litera-ture review, they can stand alone in many othercases.

Step four: synthesize and write

The importance of synthesis is best understood incontrast to its opposite, which I often term the bookreport style of writing. In book report style onereports the studies individually, indicating Jones saidthis and Smith found that, but expecting readers tointegrate the findings and draw conclusions. In bookreport style, sentences often start with the author’sname. For example, “Jones studied 15 children ages8–10.” Synthesized findings more often begin witha statement that combines or integrates severalfindings followed by the relevant references. Forexample, “Of the studies reviewed, 12 involved chil-dren ages 3–10 in sample sizes ranging from 15–220children (references).”

Synthesis is very difficult if you attempt to store allthe information in your mind at once. Using thesteps described here, however, it is relatively simpleto create an integration and summary of the findingsbecause you have recorded in evidence tables and

R. L. Foster Editorial

87Journal for Specialists in Pediatric Nursing 18 (2013) 85–88© 2013, Wiley Periodicals, Inc.

Page 4: Extracting and synthesizing information from a literature review

buckets all the relevant information. For example, inreporting the results of their systematic review, Neuand colleagues examined their evidence table towrite, “Studies (n = 4) of nonpharmacologic therapyinvestigated benefits of formulas, feeding modifica-tions, or dietary supplements” (2012, p. 186).Review of other columns in the table allowed themto complete the paragraph detailing nonpharmaco-logic studies. Information buckets could have helpedthem in writing the background, discussion, andconclusion sections of the article.

CONCLUSION

Literature reviews that solve clinical problemsand/or advance our understanding of nursing phe-nomena are critical to evidence-based practice andresearch. A simple method for identifying, extract-ing, and synthesizing the information can result in

credible answers to search/research questions.Whether the extraction and synthesis methoddescribed here is achieved individually or by agroup, the method provides a basic framework thatcan easily be modified to enhance existing schemas.

Roxie L. Foster, PhD, RN, [email protected]

References

Benner, P. (1984). From novice to expert. Excellence and powerin clinical nursing practice. Menlo Park, CA:Addison-Wesley.

Neu, M., Corwin, E., Lareau, S. C., & Marcheggiani-Howard, C. (2012). A review of nonsurgical treatmentfor the symptom of irritability in infants with GERD.Journal for Specialists in Pediatric Nursing, 17, 177–192.doi:10.1111/j.1744-6155.2011.00310.x

Editorial R. L. Foster

88 Journal for Specialists in Pediatric Nursing 18 (2013) 85–88© 2013, Wiley Periodicals, Inc.