• Tidak ada hasil yang ditemukan

Searching Bibliographic Databases

Dalam dokumen Essentials of Nursing Research (Halaman 190-195)

Print-based bibliographic resources that must be searched manually are becoming outmoded. Reviewers typically begin by searching bibliographic databases that can be accessed by computer. Several commercial vendors (e.g., Ovid, EbscoHost, Pro- Quest) offer information retrieval services for bibliographic databases. Their pro- grams are user friendly, offering menu-driven systems with on-screen support so that retrieval can usually proceed with minimal instruction. Some providers offer discount rates for students and trial services that allow you to test them before sub- scribing. In most cases, however, your university or hospital library is already a sub- scriber.

Getting Started with an Electronic Search

Before undertaking a search of a bibliographic database electronically, you should become familiar with the features of the software you are using to access the data- base. The software will give you options for restricting or expanding your search, for combining the results of two searches, for saving your search, and so on. Most programs have tutorials that offer useful information to improve the efficiency and effectiveness of your search.

An early task in doing an electronic search is to identify keywords to launch the search. A keyword is a word or phrase that captures the key concepts in your ques- tion. For quantitative studies, the keywords you begin with are usually the primary independent or dependent variables, and perhaps the population. For qualitative studies, they keywords would be the central phenomenon of interest and the popu- lation. If you have used the question templates for asking clinical questions that were included in Box 2.1, the words you entered in the blanks are likely to be good key words.

For most bibliographic database searches, there are various types of search approaches. All citations included in a database have to be indexed so they can be retrieved, and databases and programs use their own system of categorizing the entries. The indexing systems have specific subject headings(subject codes) and a hierarchical organizational structure that permits various more specific subheadings.

You can undertake a subject search by entering a subject heading into the search field. You do not have to worry, however, if you are not familiar with the spe- cific subject codes used in the program. Most software you are likely to use has mapping capabilities. Mappingis a feature that allows you to search for topics using

TIP

If you want to identify all major research reports on a topic, you need to be flexible and to think broadly about the keywords that could be related to your topic. For example, if you are interested in anorexia nervosa, you might look under anorexia,eating disorders, and weight loss, and perhaps under appetite,eating behavior,food habits, bulimia, and body weight changes.

your own keywords, rather than needing to enter a term that is exactly the same as a subject heading in the database. The software translates (“maps”) the keywords you enter into the most plausible subject heading. Then it will retrieve all citations that have been coded with that subject heading.

Even when there are mapping capabilities, you should learn the relevant sub- ject headings of the database you are using because keyword searches and subject heading searches yield overlapping but nonidentical search results. Subject head- ings for databases can be accessed in the database’s thesaurus or other reference tools. Moreover, you may get additional suggestions for how to search if you learn something about the database’s subject headings and structure.

When you enter a keyword into the search field, it is likely that the program will institute both a subject search, as just described, and a textword search. A textword search will search for your specific keyword in the text fields of the records in the database (e.g., the title or the abstract). Thus, if you searched for lung cancerin the MEDLINE database (which we describe in a subsequent section), the search would retrieve citations coded for the subject code of lung neoplasms (the MEDLINE sub- ject heading used to code citations), and also any citations in which the phrase lung cancer appeared, even if it had not been coded for the lung neoplasmsubject heading.

In addition to subject searches and textword searches, it is also possible to search for citations for a specific author. An author search might be productive if you are familiar with the names of leading researchers in a field, for example.

Although it is beyond the scope of this book to provide extensive guidance on doing an electronic bibliographic search, we can offer a few suggestions that could improve your search results. One tool available in most databases is wildcard char- acters. A wildcard character—which is a symbol such as “$” or “*”, depending on the database and search program—can be used to search for multiple words that share the same root. To do this, the wildcard character is inserted immediately after the truncated root. For example, if we entered nurs* in the search field for a MED- LINE search, the computer would search for any word that begins with “nurs,”

such as nurse, nurses, and nursing. This can be very efficient, but note that the use of a wildcard character is likely to turn off the mapping feature, and result in a textword search exclusively.

Another way to force a textword search is to use quotation marks around a phrase, which will yield citations in which the exact phrase appears in text fields.

In other words, lung cancer and “lung cancer” might yield different results. One strategy for a thorough search might be to do a search with and without wildcard characters and with and without quotation marks.

Boolean operatorsare another useful tool that can be used to expand or restrict a search. For example if you were interested in citations that concerned lung can- cer and smoking, you could enter the following: lung cancer AND smoking. The Boolean operator “AND” would instruct the computer to restrict the search to cita- tions that had both lung cancer and smoking as textwords or subject headings. The Boolean operator “OR” would expand the search—if you entered lung cancer OR smoking, you would get citations to all references with either term.

Two especially useful electronic databases for nurses are CINAHL (Cumulative Index to Nursing and Allied Health Literature) and MEDLINE (Medical Literature

On-Line), which we discuss in greater detail in the next sections. Other potentially useful bibliographic databases for nurses include the Cochrane Database of Sys- tematic Reviews, the Institute for Scientific Information (ISI)’s Web of Knowledge, and EMBASE (the Excerpta Medica database). The Web of Knowledge database is particularly effective for using the descendancy approach as a search strategy because of its strong citation indexes.

The CINAHL Database

The CINAHL databaseis an extremely important electronic database for nurses.

It covers references to virtually all English-language nursing and allied health journals, as well as to books, book chapters, dissertations, and selected conference proceedings.

The CINAHL database includes materials dating from 1982 to the present and contains more than 1.3 million records. CINAHL provides bibliographic informa- tion for locating references (i.e., the author, title, journal, year of publication, vol- ume, and page numbers), as well as abstracts of most citations. Supplementary information, such as names of data collection instruments, is available for many records in the database. Documents of interest can typically be ordered electroni- cally. CINAHL can be accessed online or by CD-ROM, either directly through CINAHL (www.cinahl.com) or through a commercial vendor. We illustrate some of the features of the CINAHL database using Ovid software, but note that some of the features discussed may be unavailable or labeled differently if you are using differ- ent software to access CINAHL.

At the outset, you might begin with a “basic search” by simply entering key- words or phrases relevant to your primary question. You may want to restrict your search in a number of ways, for example, by limiting the records retrieved to a certain type of document (e.g., only research reports); to records with certain features (e.g., only ones with abstracts); to specific publication dates (e.g., only those after 1999); or to those written in English. A few simple clicks in the field labeled “Limits” can accomplish this, and clicking on “More Limits” offers more options.

To illustrate how searches can be delimited with a concrete example, suppose we were interested in recent nursing research on children’s pain. Here is an exam- ple of how many “hits” there were on successive restrictions to the search, using one of many possible keyword search strategies, and using the CINAHL database through April 1, 2008:

SEARCH TOPIC/RESTRICTION HITS

Pain 63,372

Pain AND child$ 6,014

Limit to research reports 2,873

Limit to English 2,751

Limit to entries with abstracts 2,334

Limit to nursing journals 595

Limit to 2003 through 2008 publications 251

This narrowing of the search—from over 63,000 initial references on pain to 251 references for research reports published between 2003 and 2008 in nursing journals that has content on pain and children—took under 1 minute to perform.

Note that we used the Boolean operator “AND” to search only for records that had both keywords in which we were interested. Also, we used a wildcard character, which in Ovid is “$.” This instructed the computer to search for any word that begins with “child” such as children or childhood.

The 251 references from the search would be displayed on the monitor, and we could then print full information for ones that seemed promising. An example of an abridged CINAHL record entry for a report identified through this search on children’s pain is presented in Figure 7.2. Each entry shows an accession number that is the unique identifier for each record in the database, as well as other identi- fying numbers. Then, the authors and title are displayed, followed by source infor- mation. The source indicates the following:

* Name of the journal (Pain Management Nursing)

* Year and month of publication (2007 Dec)

* Volume (8)

* Issue (4)

* Page numbers (156–65)

* Number of cited references (33)

Figure 7.2 also shows the CINAHL subject headings that were coded for this particular study. Any of these headings could have been used in the search to retrieve this reference. Note that the subject headings include substantive/topical headings such as Comfort andPain, and also include methodologic headings (e.g., Questionnaires, Diaries) and sample characteristic headings (e.g., Child). Next, when formal instruments are used in a study, their names are printed under Instru- mentation. The abstract for the study is then presented. Additional information, not shown here, includes any funding for the study and the full list of citations.

Based on the abstract, we would then decide whether this reference was perti- nent to our inquiry. Documents referenced in the database usually can be ordered or directly downloaded, so it is not necessary for your library to subscribe to the referenced journal.

The MEDLINE Database

The MEDLINE database was developed by the U.S. National Library of Medicine (NLM), and is widely recognized as the premier source for bibliographic coverage of the biomedical literature. MEDLINE covers about 5,000 medical, nursing, and health journals published in about 70 countries and contains more than 16 million records dating back to the mid 1960s. Abstracts of reviews from the Cochrane Col- laboration are also available through MEDLINE.

The MEDLINE database can be accessed online through a commercial vendor for a fee, but this database can be accessed for free on the Internet through the PubMedwebsite (http://www.ncbi.nlm.nih.gov/PubMed). This means that anyone, anywhere in the world with Internet access can search for journal articles, and thus

PubMed is a lifelong resource regardless of your institutional affiliation. PubMed has an excellent tutorial.

MEDLINE/PubMed uses a controlled vocabulary called MeSH(Medical Sub- ject Headings) to index articles. MeSH terminology provides a consistent way to retrieve information that may use different terminology for the same concepts. If you begin a search with your own keyword, you can click on the “Display” tab near

ACCESSION NUMBER

2009743835 NLM Unique Identifier: 18036503.

AUTHOR

Wiggins SA, Foster RL INSTITUTION

University of Nebraska Medical Center College of Nursing, Lincoln, Nebraska.

TITLE

Pain after tonsillectomy and adenoidectomy: "ouch it did hurt bad".

SOURCE

Pain Management Nursing. 2007 Dec; 8(4): 156-65. (33 ref) ABBREVIATED SOURCE

PAIN MANAGE NURS. 2007 Dec; 8(4): 156-65. (33 ref) CINAHL SUBJECT HEADINGS

*Adenoidectomy Content Analysis Midwestern United States

Adolescence Convenience Sample Nausea and Vomiting

Age Factors Descriptive Research *Pain Measurement

Analgesics/ad [Admin, Dosage] Descriptive Statistics Patient Attitudes/ev [Evaluation]

Analysis of Variance Diaries *Postoperative Pain

Art Female Purposive Sample

Audiorecording Friedman Test Questionnaires

Chi Square Test Hospitals Record Review

Child Interviews Repeated Measures

Comfort Male Thematic Analysis

Constant Comparative Method Medical Records *Tonsillectomy

INSTRUMENTATION

Hurt/Pain Diary. Body Outline for children aged 4–7 years (Van Cleve and Savedra). Poker Chip Tool (PCT) (Hester).

ABSTRACT

Severe pain experiences for children at home after tonsillectomy and adenoidectomy (T & A) have been described for more than a decade. Children and their parents are responsible for pain and symptom management during the postoperative home recovery. The purpose of this research was to more fully explore the pain experience and home management practices from the child's perspective. Diaries were used by 34 children (4–18 years of age) to docu- ment their pain and other symptoms. A home interview, stories, art work, and personal notes were also explored to fully capture the experience. From the evening of surgery through the second postoperative day, children reported mean pain intensity ratings of 3.1–3.3 out of a possible 4. Pain awakened 64.7% of the children from nighttime sleep and 52.9% reported vomiting associated with nausea. Children received an average of only 50% of the analgesic doses prescribed. Across the 3 postoperative days studied, pain remained severe and interventions offered minimal relief. Neither older children (chi(2) 1.357, n 13, df 2, p .259) nor younger children (chi(2) 1.357, n 12, df 2, p .507) reported significant differences in their mean pain intensity across the first 3 postoperative days.

Results supported concerns for inadequate home pain management practices in the pediatric T & A population.

Example of a printout from a CINAHL (Cumulative Index to Nursing and Allied Health Literature) search.

FIGURE 7.2

the top right of the screen to see how the term you entered mapped onto MeSH terms, which might lead you to pursue other leads. You can also search for refer- ences using the MeSH database directly by clicking on “MeSH database” in the left blue panel of the PubMed home page, listed under PubMed services. Note that MeSH subject headings may overlap with, but are not identical to, the subject head- ings used in the CINAHL database

If we did a PubMed search of MEDLINE similar to the one we described ear- lier for CINAHL, this is what we would find:

SEARCH TOPIC/RESTRICTION HITS

Pain 369,331

Pain AND child* 36,424

Limit to research reports (NA)

Limit to English 29,353

Limit to entries with abstracts 24,588

Limit to nursing journals 1,103

Limit to 2003 through 2008 publications 390

This search yielded more references than the CINAHL search, but we were not able to limit the search to research reports because the publication types in this database are categorized differently than in CINAHL. For example, in PubMed we could limit the search to certain typesof research (e.g., clinical trials), but it may be risky to use this restriction criterion. Note that in PubMed, the wildcard code used to extend truncated words is * rather than $.

Figure 7.3 shows the full citation for the same reference we located earlier in the CINAHL database (Figure 7.2). To get this full citation, you would need to use a pull-down menu near the top that offers alternative Displays—in this case, you would designate “Citation.” The citation display presents all the MeSH terms that were used for this reference and, as you can see, the MeSH terms are different than the subject headings in CINAHL.

Dalam dokumen Essentials of Nursing Research (Halaman 190-195)