Practice Matters — How to appraise quantitative research articles By Kenneth J. Rempher, PhD, RN, MBA, CCRN, APRN, BC, and Cory Silkman, MAR, BSN, RN, C WHATEVER THEIR SPECIALTY Don’t let the research process intimidate you. Use these guidelines to critically appraise scientific studies. or practice area, all nurses should strive to become more sophisticated consumers of nursing research by learning how to critically appraise, synthesize, and communicate research findings. Such critical appraisal shows your commitment to evidence-informed practice and empowers you to create a practice culture based on the best available evidence. To critically appraise nursing research, you must ask focused, meaningful questions to determine the overall integrity and applicability of the research. This article will help you better understand—and undertake—the process of critical appraisal. With practice, even nurses who once were intimidated by the research process should be able to efficiently and effectively determine the clinical relevance of scientific studies. (See Questions to ask yourself when evaluating a scientific study.) Problem statement problem statement can cause flaws in a study’s methods, protocols, samples, and analyses. Be wary of broadly stated or overly generalized problem statements, as well as those whose research questions can’t be answered by the methods proposed. (See Parsing the problem statement.) Literature review The literature review is a systematic, critical review of the most important scholarly literature on a given topic. It should: • highlight critical weaknesses in previous studies • identify previously studied concepts or variables • relate the current research project to historical research • identify the current knowledge deficit about a particular phenomenon and state what more needs to be done to overcome that deficit. Look for a broad range of references—for example, peer-reviewed journal articles, systematic reviews of relevant research, professional standards, position statements, dissertations, and conference proceedings. In general, references should be no more than 5 years old, unless the research cited is a classic or historically important work. The problem statement should appear at the beginning of the article and should include enough information for you to determine if the study results can be generalized to a specific patient population. Effective problem statements include independent and dependent variables, population of interest, and key concepts of the study. The research problem should provide a clear rationale for the study. The problem statement may be in To better illustrate the importance of a clearly written problem statement, conone of two forms: sider these two examples. • a research question that indicates the who, what, when, where, and #1: “The primary purpose of this study was to determine the influence of clinical why of the study pathways on heart failure outcomes.” • a “purpose” statement that de#2: “The primary purpose of this study was to assess the impact of a nurse-driven scribes the researcher’s purpose in clinical pathway on length of stay, costs, charges per patient day, documentation of smoking cessation advisement, and administration of ACE inhibitors for conducting the study. Medicare patients diagnosed with left ventricular systolic dysfunction.” Also, the research problem should fill a gap in the current body of nursing Statement #1 is broad and general and doesn’t identify the independent and deresearch or theory or should pinpoint a pendent variables, population of interest, and key concepts being studied. Statesingle, relevant nursing issue that’s ment #2 provides all relevant information, including dependent and independmeaningful to nurses and patients. ent variables, population of interest, and key concepts. A poorly worded or inappropriate Parsing the problem statement 26 American Nurse Today January 2007 Conceptual framework Depending on the nature of the study, a conceptual or theoretical framework may be presented near the beginning of the article. • Theoretical frameworks are narrower in scope and can be tested directly. • Conceptual frameworks express assumptions and can’t be tested directly. A well-defined conceptual framework allows the reader to better understand the relationship between major concepts of the study and more fully explicates the relationship between the variables. Additionally, a conceptual framework may help you better understand how the researcher’s hypothesis and research question were developed. Methods The methods section of the article tells you how the principal investigator went about answering the research question. It includes information about the sample selection, study design, data collection, and data or statistical analysis. This section should also provide sufficient information to permit duplication of the study and should address the protection of human subjects. Sample selection Sample selection occurs on the basis of eligibility criteria that the researcher establishes in accordance with the study’s objectives. The sample’s eligibility and exclusion criteria, as well as its demographic composition, should be appropriate to achieve the study’s objectives. The study should have an adequate number of participants and a low dropout rate to protect against compositional and statistical bias and make the study more representative of the population. Study design The study design should be clearly stated and appropriate for the research question being asked. The most common design associated with quantitative research is experimental design—commonly considered the most rigorous design and, for many researchers, the gold standard. In this design, the researcher controls both the selection of study subjects and introduction of the independent variable. Also, in experimental design, subjects are randomly Questions to ask yourself when evaluating a scientific study Use this checklist to help you determine if a study is valid and applicable. Problem statement or research question ■ Does the problem statement include the population of interest? ■ Does it identify key study concepts? ■ Does the research question fill a gap in the present body of nursing knowledge or theory? ■ Does the question clearly define the relationship between the variables? ■ Does the question address the who, what, where, when, and why of the research study? Literature review Does the literature review: ■ highlight weaknesses of previous studies? ■ relate the current study to historical research? ■ address the study variables and hypotheses? ■ make evident the current deficit in knowledge about the phenomenon of interest? ■ state what needs to be done to overcome the existing deficit? ■ cover a broad array of resources? ■ cite studies no older than 5 years? Conceptual or theoretical framework ■ Is the framework described? ■ Does it identify the relationship between the variables? ■ Does it support the study hypothesis? Sample selection ■ What method was used to select the sample? ■ Do the eligibility and exclusion criteria help achieve the study’s objectives? ■ Is the sample size appropriate and representative of the intended population? ■ Is the dropout rate significant enough to skew results? Methods and design ■ Are all instruments used to collect data adequately described? ■ Is there evidence of validity and reliability for the instruments? ■ Is the design appropriate for the study? ■ Is the design sufficiently described so it could be easily replicated? ■ Are both descriptive and inferential statistics presented (if appropriate)? ■ Is the effect size presented? ■ Are the statistical tests used for analysis appropriate? Protection of human subjects ■ Did the researcher obtain informed consent from subjects or their legal representatives? ■ Did an institutional review board approve the study? ■ Do the study’s benefits outweigh its potential risks to participants? Results and discussion of findings Does the author: ■ relate statistical findings to the study’s independent and dependent variables? ■ clearly identify statistical tests and methods used to analyze the data? ■ relate study findings to the larger conceptual or theoretical framework? ■ adequately address the study’s limitations and weaknesses? January 2007 American Nurse Today 27 assigned to treatment and control groups, thereby reducing study bias because the researcher can’t influence the assignment of subjects. Ideally, groups in an experimental study Be aware that the “p” value tells you only that a difference exists between the are similar in all respects, and any experimental and control groups. It doesn’t tell you the magnitude of the effect on differences between them result from the experimental group. Understanding the magnitude of effect helps distinguish the intervention administered by the clinical significance from statistical significance. The article should report statistics researcher. Experimental design is known as the effect size and confidence intervals to help you determine the commonly used when new drugs or intervention’s magnitude of effect. medical products are being studied. In contrast, the nonexperimental Protection of human subjects study design is more qualitative and is used when the reTo protect research subjects, the study must be approved searcher wants to observe a particular phenomenon but by an institutional review board. The researcher also lacks the ability and desire to manipulate the independent must obtain informed consent from participants after givvariables. The quasi-experimental design is more closely ing them oral and written information on the nature of related to experimental design but lacks random assignthe study and potential safety risks or conflicts of interest. ment of subjects and subsequently may introduce bias. What the “p” value doesn’t tell you Data collection Data collection procedures should be fully explained in the methods section and should provide a clear understanding of how data were collected and who collected them. Issues such as inter-rater reliability, instrument reliability and validity, and training of data collectors should be addressed. A clear explanation of data collection lends credibility to the study. Statistical analysis A thorough review of the statistical analysis is important. (If you’re uncomfortable or unfamiliar with basic statistics, consult a clinical nurse specialist, nurse researcher, or other advanced practice nurse for assistance.) Study results should be presented in a logical, systematic format. For quantitative studies, both descriptive and inferential statistics should be provided. • Descriptive statistics, including measures of central tendency (mean, median, and mode) as well as measures of dispersion or variability (variance, standard deviation, and range) provide information about the characteristics of the subjects or phenomenon being studied. • Inferential statistics allow researchers to make assumptions about the population based on a sample. Through the use of significance tests and other measures, inferential statistics help researchers understand the probability that the results of their study occurred by chance. The level of significance, expressed as a “p” value, represents the probability of obtaining the computed value by chance. If “p” is less than 0.01, the probability of obtaining the computed results by chance is less than 1%. In other words, if the study were repeated 100 times, the difference between groups would be attributed to the study intervention 99 times and to chance only one time. (See What the “p” value doesn’t tell you.) 28 American Nurse Today January 2007 Results and discussion of findings Here the author should succinctly relate any descriptive and inferential statistical findings to the study’s independent and dependent variables (as stated in the original research question or research hypothesis). This section should state clearly whether the data analysis supports, or fails to support, the research hypothesis. You should be able to read the study’s results and determine the relationship between the independent and dependent variables. Some articles may contain a separate “discussion” section in which the author interprets, analyzes, and summarizes the study’s conclusions and its relevance to the larger theoretical framework. Expect the author to objectively state any limitations or weaknesses in the study’s design, method, sample, or data collection procedures. The discussion also should identify any conceptual or theoretical relationships in need of further investigation. ✯ Selected references Dunning M, Abi-Aad G, Gilbert D. Experience, Evidence and Everyday Practice: Creating Systems for Delivering Effective Health Care. London: King's Fund; 1999. Greenhalgh T. How to read a paper. The Medline database. BMJ. 1997;315(7101):180-183. Griffin-Sobel JP. Research in practice: immersing yourself in research. Gastroenterol Nurs. 2003;26(5):219-220. Hudson-Barr D. From research idea to research study: the how. J Spec Pediatr Nurs. 2005;10(3):147-150. LoBiondo-Wood J, Haber J. Nursing Research: Methods, Critical Appraisal, and Utilization. St. Louis, Mo: C.V. Mosby; 2002. Valente S. Critical analysis of research papers. J Nurs Staff Dev. 2003;19(3):130-142. Kenneth J. Rempher, PhD, RN, MBA, CCRN, APRN, BC, is Director of Professional Nursing Practice at Sinai Hospital of Baltimore in Baltimore, Md. Cory Silkman, MAR, BSN, RN, C, is a Clinical Leader in the Comprehensive Inpatient Rehabilitation Unit at Sinai Hospital of Baltimore.