# Applying Statistics to Clinical Nurse Specialist

## Applying Statistics to Clinical Nurse Specialist

Applying Statistics to Clinical Nurse Specialist Applying Statistics to Clinical Nurse Specialist Click to read the article Applying Statistics to Clinical Nurse Specialist Practice , and discuss the relationship between descriptive and inferential statistics, including their related applications. Present an issue in your field of practice of nursing and a potential application of statistics to it.Applying Statistics to Clinical Nurse Specialist Practice Larson, R. and Farber, B. (2011). Elementary Statistics: Picturing the World (5 th ed.) Upper Saddle River, NJ: Pearson Prentice Hall. Chapter 8, Sections 8.1-8.3, and Chapter 10, Section 10.4 MUST BE APA FORMAT MUST BE 250-300 WORDS ORDER NOW FOR CUSTOMIZED AND ORIGINAL ESSAY PAPERS applying_stats.pdf Applying Statistics to Clinical Nurse Specialist. Clinical Nurse SpecialistA Copyright B 2012 Wolters Kluwer Health | Lippincott Williams & Wilkins Using Research to Advance Nursing Practice Column Editor: Janice Buelow, PhD, RN, FAAN Applying Statistics to Clinical Nurse Specialist Practice Cheryl Westlake, PhD, RN, ACNS-BC n Renee Pozza, PhD, RN, CNS, FNP-BC F lorence Nightingale was the first known nurse statistician. Her use of statistical analysis techniques regarding the incidence of preventable military death caused by unsanitary conditions clearly substantiated the need for reform. With her analysis, she revolutionized the idea that social phenomena could be objectively measured and subjected to mathematical analysis. Clinical nurse specialists (CNSs), too, must be able to analyze clinical phenomena in a systematic manner in order to use and generate research in the clinical setting for enhanced patient care. This article addresses the purpose and types of statistics, research questions, levels of measurement, and the selection, application, and evaluation of statistical analysis techniques. A case study format is used to represent the content and facilitate application of the information. However, the authors recognize that this article does not replace a statistics class or a reference text, but provides a userfriendly primer of statistics for the clinical nurse specialist (CNS). Case Study A CNS wishes to study the relationship between a CNS consultation intervention regarding palliative care for patients with heart failure (HF) and patient outcomes of spirituality, quality of life (QOL), and emergency room (ER) visits postdischarge. Furthermore, the CNS would like to determine the difference between patients who receive the CNS consultation intervention and those who did not. As part of the report to administration, the CNS would like to describe the patient population and the average level of spirituality, QOL, and the average number of ER visits postdischarge. Author Affiliations: Professor and Associate Dean, International and Community Programs (Dr Westlake), Associate Professor and Associate Dean, Academic Affairs (Dr Pozza), School of Nursing, Azusa Pacific University, California. The authors report no conflicts of interest. Correspondence: Cheryl Westlake, PhD, RN, ACNS-BC, International and Community Programs, WC 311, 720 E Foothill Blvd, School of Nursing, Azusa Pacific University, Azusa, CA 91702 ([email protected]). DOI: 10.1097/NUR.0b013e318253624c Clinical Nurse Specialist A OVERVIEW Statistics are merely applied mathematics that deal with data collection, organization, and interpretation using well-defined procedures.1 The purposes of statistics are to: n describe and summarize information; n identify associations, relationships, or differences; and n facilitate predictions or generalizations. There are two broad types of statistics: descriptives; and inferential. Descriptive statistics describe or characterize data by summarizing them into more understandable terms. Summary tables, charts, frequencies, percentages, and measures of central tendency are frequently used descriptive statistics. Inferential statistics are techniques used that provide the ability to infer or predict population characteristics from a sample or subset of the population.Applying Statistics to Clinical Nurse Specialist Practice Applying Statistics to Clinical Nurse Specialist. In nursing, inferential statistics are frequently used to compare two or more groups in an attempt to determine if the two groups are different from one another or if an intervention is better than usual care or another intervention. The CNS may wish to generalize the findings of the study sample to all patients. Examples of inferential statistics include t tests, correlations, # 2, and regressions. Case Study The CNS in our case study will use both descriptive and inferential statistics. The descriptive statistics will be used to describe the patient population demographics and the groups average level of spirituality and QOL, and the average number of ER visits postdischarge. The data from each patient in the sample will be aggregated and summarized into more understandable terms. The data may be represented by frequencies, percentages, and measures of central tendency. As the CNS in our case study wishes to (1) determine if the CNS consultation intervention is better than usual care and (2) study the relationship between the CNS consultation intervention and patient outcomes, inferential statistics will be used in addition to the descriptive data. The CNS may wish to generalize the findings of the CNS consultation intervention for patients in his/her sample to all patients with HF. In classic inferential statistics, two hypotheses are made before the study commences: the null hypothesis; and www.cns-journal.com Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 131 Using Research to Advance Nursing Practice the alternative hypothesis. The null hypothesis states that the two groups studied will be the same, ie there will be no significant difference between the groups. The alternative hypothesis states that the two groups studied will be different. Thus, the goal in classic, inferential statistics is to prove the null hypothesis wrong. In order to determine the type of and specific statistical test needed, the CNS will consider the research question and identify the variables. Is the study about a single variable of interest or multiple variables? If there are multiple variables, are the variables dependent on the outcome of the other variable(s) or intervention? If yes, then which variables are dependent on the outcome variable(s)? An independent variable (IV) is presumed to cause or explain or account for variations in the dependent variable (DV). For CNSs, the IV is the factor that is manipulated or the intervention that results in the change in the DV, or outcome. For example, in the current clinical trial listed on ClinicalTrials.gov by Keith2 (http://clinicaltrials.gov/ct2/show/ NCT00959075 accessed 2/13/2012), the purpose is to determine if providing oral thiamin supplementation improves heart function, symptoms, exercise capacity, and/or QOL in patients with HF. Oral thiamin supplementation is the intervention or IV, and left ventricular ejection fraction is the primary outcome, or DV. Important to note, the IV may not always be manipulated or changed by the CNS. Often, the CNS attempts to determine if a factor present in the clinical environment results in a change in an important clinical outcome. For example, Albert and colleagues3 analyzed fungal and bacterial growth on cleaned, ready-to-use, reusable electrocardiographic lead wires to determine if an association existed between the hospital site or work environment. Fungal and bacterial growth was the DV or outcome variable and hospital site or work environment was the IV. Furthermore, social interventions may be examined by the CNS. A study by Rayens and colleagues4 evaluated the effect of a smoke-free law on the rate of emergency department visits for asthmaApplying Statistics to Clinical Nurse Specialist Practice Applying Statistics to Clinical Nurse Specialist. The smoke-free law was the intervention or IV, and ER visits for asthma was the outcome or DV. Common to all studies is the need to determine the IV and the DV based on the research question(s) with clear definitions and measurement criteria. RESEARCH QUESTION AND HYPOTHESES The research question in this case could be: What difference in level of spirituality, QOL, and number of ER visits postdischarge, if any, exists between one group of patients with HF who receive standard care compared with one group of patients with HF who receive a CNS consultation intervention regarding palliative care? Null hypothesis: No statistically significant differences will be observed between groups on self-reported spirituality, QOL, and number of postdischarge ER visits. Alternative hypothesis: Statistically significant differences will be observed between groups on self-reported spirituality, QOL, and number of postdischarge ER visits. MEASUREMENT ISSUES Case Study The IV in the CNS case study is the CNS consultation intervention regarding palliative care for patients with HF. The DVs are the patient outcomes of spirituality, QOL, and ER visits postdischarge. Once the CNS has described the study variable(s), levels of measurement for each of the variables must be determined. A helpful acronym for recalling the levels in ascending order is NOIR: nominal; ordinal; interval; and ratio. The levels of measurement describe the type of data obtained. Nominal data include categorical dataVdata that are in named (nominal) categories: yes/no, present/absent, male/female, and group assignment. Nominal data may include more than two categories such as group assignment to experimental, control, or standard-of-care group. Ordinal data also include categorical data, but unlike nominal data where the groups differ in name but not in order, ordinal data place an order to the categories. The distance between categories is unknown, but the order matters. Examples of ordinal data may include (1) socioeconomic status: categorized as low, middle, or high or (2) pain: characterized as absent, minimal, mild, moderate, or severe. Interval data are ordinal data where the distance between categories is of equal intervals and there is a true zero. Examples of interval data include temperature and pain using a 0- to 10-point scale. Finally, ratio data refer to meaningfully ordered categories with equal intervals between them and a true 0 point. Examples of ratio data include blood pressure, heart rate, and weight. Case Study Case Study The hypotheses generated by the CNS will be dependent on the literature review and supporting data. Suppose that previous studies address the concepts of spirituality, QOL, and number of ER visits in patients with HF,5Y14 but only QOL has been addressed as a DV in a study of a CNS intervention in patients with HF.10 However, spirituality, QOL, and ER visits postdischarge could serve as DVs in our case study. QOL would have data to support a hypothesis. QOL, and ER visits postdischarge. The CNS determines that the concepts will be measured using the instruments and data type as indicated below. Then, the CNS determines how many patients are needed in each group (intervention versus standard of care). 132 Concept CNS palliative care consultation intervention Measure Data Type Investigator-developed item (yes/no) Nominal www.cns-journal.com Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. May/June 2012 Spirituality Spirituality (visual analog scale 0Y10)15 Interval QOL Short-Form 1216 Ordinal ER visits postdischarge Investigator-developed item (no. of visits) Ratio In order to determine the number of patients or subjects needed in each group, a power analysis may be performed. Power is the ability to detect real differences or relationships that exist in the population from which the sample is drawn. Power is calculated using a mathematical formula based on the degree of significance, the expected effect size, and the desired level of power.17Y18 There are online calculators that compute the minimum required sample size using the P value, number of IVs, anticipated effect size (0.02, 0.15, and 0.35 and small, medium, and large, respectively), and the desired statistical power level (usually Q 0.8). One such calculator that may be useful to CNSs is the free A-priori Sample Size Calculator for Multiple Regression from Daniel Sopers Free Statistics Calculators web site available at http://danielsoper.com/statcalc3/calc.aspx?id = 1. Now that the needed sample size is known, the CNS is ready to obtain the appropriate institutional review board approvals and begin data collection. When the CNS begins data analysis, the first step is to plot the data in order to note the distribution and mode of the data. The next step is to determine whether the distribution is normal or not. Applying Statistics to Clinical Nurse Specialist Practice Applying Statistics to Clinical Nurse Specialist . A normal distribution is characterized by a unimodal, symmetrical, bell-shaped curve, when interval data are represented by a histogram or line graph.19(p128) Normal distribution may be determined by visually inspecting the histogram or line graph and analyzing the similarity of the mean and median. The closer to equal the mean and the median are, the closer to a normal distribution the data are.19 The second step is to determine the critical significance level, specifically the risk the CNS is willing to take about being incorrect. The critical significance level is represented as the alpha and is the probability that the relationship between or difference in means between groups is not happening by chance. Typically, the ! level is set at .05, which means that if P is less than .05, those differences are not due to chance. The significance level enables judgment of the evidence against the null hypothesis or evidence for the alternative hypothesis. The P value ranges from 0.00 to 1.00, with a large P value supporting the null hypothesis (there is no difference) or a small P value supporting the alternative hypothesis (there is a difference). Usually, an significance level of .05 is acceptable unless the risk of a type I error, false positive, or an incorrect decision is great. In the case of high error risk, such as a clinical drug trial, then a significance level of .01 is used. A Clinical Nurse Specialist A significance level of .05 means that one can be 95% confident in our findings or have a 5% chance of error. A significance level of .01 means one can be 99% confident in our findings, or have a 1% chance of error. Case Study After having determined the distribution and mode of the data, the CNS would need to ascertain whether the distribution is normal or not. Then, the CNS would need to decide the acceptable significance level. Next, the issue of a one-tailed or a two-tailed test must be addressed. A one-tailed test seeks either an increase or a decrease in the measured concept and is used when the research hypothesis states direction. A two-tailed test seeks any change (either an increase or a decrease) in the concept and is used when sufficient knowledge is not available to predict direction in the measured concept. Case Study In the proposed CNS case study, limited previous knowledge facilitates limited prediction of the direction of the measured concepts. Thus, a two-tailed test will be used in the statistical analysis. The next task in statistical analysis is to determine the appropriate analysis. First, the level of the research, if determined, may facilitate selection of the appropriate statistical test. There are three levels of quantitative research: descriptive, comparative/correlational, and quasi-experimental/ experimental. The research levels and the associates statistical tests are described in ascending order in Tables 1 and 2. If a level II or III study is conducted, then additional questions about the data must be considered in order to determine the appropriate statistical test. Are the data normally distributed? Are the data independent? Is the sample variance equal to the population variance? Are the data measured on an interval or ratio scale, and are Table 1. Level 1 Research Data Level Type of Research Analysis Level I Descriptive Descriptive statistics Nominal or Ordinal Interval or Ratio Frequency: number of observations in each category Mean: sum of values divided by the number of values Displayed as frequency tables, histograms, graphs, figures Median: middle value of a set of ordered numbers; mode: most frequent value or category in a distribution www.cns-journal.com Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 133 Using Research to Advance Nursing Practice Table 2. Levels II and III Research Levels Type of Research Analysis Parametric Nonparametric Level II Comparative: differences between means Level I plus inferential statistics t Test: 2 groups; interval or ratio data Mann-Whitney (U) test: 2Applying Statistics to Clinical Nurse Specialist Practice Applying Statistics to Clinical Nurse Specialist .random independent samples; at least ordinal scale (a) 2 Independent groups: independent t test Kruskal-Wallis test (K): Q2 unmatched (unrelated) groups; nominal, ordinal, or interval scale; if only 3 groups, must be Q5 observations in each group (b) 2 Matched (related) groups: paired t test Wilcoxon test (W): 2 matched (related) groups; interval scale Analysis of variance (df, F value, and P level): Q2 unmatched (unrelated) groups Correlational Level I plus inferential statistics Univariate # 2 Test: nominal scale Pearson R: an interval scale Spearman D : ordinal, interval or ratio scale Multivariate Kendall I : ordinal scale Regression: continuous variables Level III Quasi-experimental Experimental All available statistical tests and/or analyses Path analysis (causal modeling) Confirmatory factor analysis Structural regression models the data continuous? If the answer to all of these questions is yes, then parametric analyses may be used. If the answer to any of these questions is no, then nonparametric analyses must be used. With nominal or ordinal data, nonparametric statistics are commonly used. The selection of the appropriate statistical test is based the number of the groups being analyzed, and the data scale, also (Tables 1 and 2). A helpful interactive site that facilitates selection of the correct statistical test needed is available at http://statistics.laerd.com/selecting-statistical-tests/. Once the appropriate test is determined, the analyses are conducted, and the statistics are assessed for statistical significance. If P is less than the predetermined critical ! level (P G !), the probability is small that difference or relationship occurred by chance. Thus, the null hypothesis is rejected and the alternative hypothesis accepted. If P is more than critical ! level (P 9 !), the probability is high that the difference or relationship occurred by chance. Thus, the null hypothesis is accepted and the alternative hypothesis is rejected. Thus, in summary, statistics are tools in the armament of the CNS to facilitate the answering of important clinical questions. Therefore, knowledge of statistics is essential for the CNS as this knowledge enables the evaluation of research by others and guides the research work of the CNS. The findings of such research describe the data in 134 Case Study The research questions in the CNS case study seek to describe the level of the DVs. Assuming the data are normally distributed, the needed statistical analysis is described below. Concept Data Type Statistical Analyses Spirituality Interval Mean, median, mode QOL Ordinal Mean, median, mode of each subscale score ER visits postdischarge Ratio Mean, median, mode Spirituality Interval Pearson R QOL Ordinal Spearman D ER visits postdischarge Ratio Pearson R The CNS seeks to compare patients who receive the CNS consultation intervention regarding palliative care with those who receive usual care. Assuming the data are normally distributed and the two groups are independent. Concept Data Type Statistical Analyses CNS palliative care consultation intervention Nominal Frequency, percent Spirituality Interval Independent t test QOL Ordinal Mann-Whitney U ER visits postdischarge Ratio Independent t test www.cns-journal.com Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. May/June 2012 meaningful terms, contribute to the disciplines knowledge base, and augment our understanding of clinical issues. References 1. Hazard Munro B. Statistical Methods for Health Care Research, Volume 2. Philadelphia, PA: Lippincott Williams & Wilkins; 2005. 2. Keith ME. A trial of thiamin supplementation in patients with heart failure. 2011. http://clinicaltrials.gov/ct2/show/NCT00959075. Accessed February 21, 2012. 3. Albert NM, Hancock K, Murray T, et al. Cleaned, Applying Statistics to Clinical Nurse Specialist Practice Get a 10 % discount on an order above $ 100 Use the following coupon code : NURSING10

## Applying Statistics to Clinical Nurse Specialist

Applying Statistics to Clinical Nurse Specialist Applying Statistics to Clinical Nurse Specialist Click to read the article Applying Statistics to Clinical Nurse Specialist Practice , and discuss the relationship between descriptive and inferential statistics, including their related applications. Present an issue in your field of practice of nursing and a potential application of statistics to it.Applying Statistics to Clinical Nurse Specialist Practice Larson, R. and Farber, B. (2011). Elementary Statistics: Picturing the World (5 th ed.) Upper Saddle River, NJ: Pearson Prentice Hall. Chapter 8, Sections 8.1-8.3, and Chapter 10, Section 10.4 MUST BE APA FORMAT MUST BE 250-300 WORDS ORDER NOW FOR CUSTOMIZED AND ORIGINAL ESSAY PAPERS applying_stats.pdf Applying Statistics to Clinical Nurse Specialist. Clinical Nurse SpecialistA Copyright B 2012 Wolters Kluwer Health | Lippincott Williams & Wilkins Using Research to Advance Nursing Practice Column Editor: Janice Buelow, PhD, RN, FAAN Applying Statistics to Clinical Nurse Specialist Practice Cheryl Westlake, PhD, RN, ACNS-BC n Renee Pozza, PhD, RN, CNS, FNP-BC F lorence Nightingale was the first known nurse statistician. Her use of statistical analysis techniques regarding the incidence of preventable military death caused by unsanitary conditions clearly substantiated the need for reform. With her analysis, she revolutionized the idea that social phenomena could be objectively measured and subjected to mathematical analysis. Clinical nurse specialists (CNSs), too, must be able to analyze clinical phenomena in a systematic manner in order to use and generate research in the clinical setting for enhanced patient care. This article addresses the purpose and types of statistics, research questions, levels of measurement, and the selection, application, and evaluation of statistical analysis techniques. A case study format is used to represent the content and facilitate application of the information. However, the authors recognize that this article does not replace a statistics class or a reference text, but provides a userfriendly primer of statistics for the clinical nurse specialist (CNS). Case Study A CNS wishes to study the relationship between a CNS consultation intervention regarding palliative care for patients with heart failure (HF) and patient outcomes of spirituality, quality of life (QOL), and emergency room (ER) visits postdischarge. Furthermore, the CNS would like to determine the difference between patients who receive the CNS consultation intervention and those who did not. As part of the report to administration, the CNS would like to describe the patient population and the average level of spirituality, QOL, and the average number of ER visits postdischarge. Author Affiliations: Professor and Associate Dean, International and Community Programs (Dr Westlake), Associate Professor and Associate Dean, Academic Affairs (Dr Pozza), School of Nursing, Azusa Pacific University, California. The authors report no conflicts of interest. Correspondence: Cheryl Westlake, PhD, RN, ACNS-BC, International and Community Programs, WC 311, 720 E Foothill Blvd, School of Nursing, Azusa Pacific University, Azusa, CA 91702 ([email protected]). DOI: 10.1097/NUR.0b013e318253624c Clinical Nurse Specialist A OVERVIEW Statistics are merely applied mathematics that deal with data collection, organization, and interpretation using well-defined procedures.1 The purposes of statistics are to: n describe and summarize information; n identify associations, relationships, or differences; and n facilitate predictions or generalizations. There are two broad types of statistics: descriptives; and inferential. Descriptive statistics describe or characterize data by summarizing them into more understandable terms. Summary tables, charts, frequencies, percentages, and measures of central tendency are frequently used descriptive statistics. Inferential statistics are techniques used that provide the ability to infer or predict population characteristics from a sample or subset of the population.Applying Statistics to Clinical Nurse Specialist Practice Applying Statistics to Clinical Nurse Specialist. In nursing, inferential statistics are frequently used to compare two or more groups in an attempt to determine if the two groups are different from one another or if an intervention is better than usual care or another intervention. The CNS may wish to generalize the findings of the study sample to all patients. Examples of inferential statistics include t tests, correlations, # 2, and regressions. Case Study The CNS in our case study will use both descriptive and inferential statistics. The descriptive statistics will be used to describe the patient population demographics and the groups average level of spirituality and QOL, and the average number of ER visits postdischarge. The data from each patient in the sample will be aggregated and summarized into more understandable terms. The data may be represented by frequencies, percentages, and measures of central tendency. As the CNS in our case study wishes to (1) determine if the CNS consultation intervention is better than usual care and (2) study the relationship between the CNS consultation intervention and patient outcomes, inferential statistics will be used in addition to the descriptive data. The CNS may wish to generalize the findings of the CNS consultation intervention for patients in his/her sample to all patients with HF. In classic inferential statistics, two hypotheses are made before the study commences: the null hypothesis; and www.cns-journal.com Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 131 Using Research to Advance Nursing Practice the alternative hypothesis. The null hypothesis states that the two groups studied will be the same, ie there will be no significant difference between the groups. The alternative hypothesis states that the two groups studied will be different. Thus, the goal in classic, inferential statistics is to prove the null hypothesis wrong. In order to determine the type of and specific statistical test needed, the CNS will consider the research question and identify the variables. Is the study about a single variable of interest or multiple variables? If there are multiple variables, are the variables dependent on the outcome of the other variable(s) or intervention? If yes, then which variables are dependent on the outcome variable(s)? An independent variable (IV) is presumed to cause or explain or account for variations in the dependent variable (DV). For CNSs, the IV is the factor that is manipulated or the intervention that results in the change in the DV, or outcome. For example, in the current clinical trial listed on ClinicalTrials.gov by Keith2 (http://clinicaltrials.gov/ct2/show/ NCT00959075 accessed 2/13/2012), the purpose is to determine if providing oral thiamin supplementation improves heart function, symptoms, exercise capacity, and/or QOL in patients with HF. Oral thiamin supplementation is the intervention or IV, and left ventricular ejection fraction is the primary outcome, or DV. Important to note, the IV may not always be manipulated or changed by the CNS. Often, the CNS attempts to determine if a factor present in the clinical environment results in a change in an important clinical outcome. For example, Albert and colleagues3 analyzed fungal and bacterial growth on cleaned, ready-to-use, reusable electrocardiographic lead wires to determine if an association existed between the hospital site or work environment. Fungal and bacterial growth was the DV or outcome variable and hospital site or work environment was the IV. Furthermore, social interventions may be examined by the CNS. A study by Rayens and colleagues4 evaluated the effect of a smoke-free law on the rate of emergency department visits for asthmaApplying Statistics to Clinical Nurse Specialist Practice Applying Statistics to Clinical Nurse Specialist. The smoke-free law was the intervention or IV, and ER visits for asthma was the outcome or DV. Common to all studies is the need to determine the IV and the DV based on the research question(s) with clear definitions and measurement criteria. RESEARCH QUESTION AND HYPOTHESES The research question in this case could be: What difference in level of spirituality, QOL, and number of ER visits postdischarge, if any, exists between one group of patients with HF who receive standard care compared with one group of patients with HF who receive a CNS consultation intervention regarding palliative care? Null hypothesis: No statistically significant differences will be observed between groups on self-reported spirituality, QOL, and number of postdischarge ER visits. Alternative hypothesis: Statistically significant differences will be observed between groups on self-reported spirituality, QOL, and number of postdischarge ER visits. MEASUREMENT ISSUES Case Study The IV in the CNS case study is the CNS consultation intervention regarding palliative care for patients with HF. The DVs are the patient outcomes of spirituality, QOL, and ER visits postdischarge. Once the CNS has described the study variable(s), levels of measurement for each of the variables must be determined. A helpful acronym for recalling the levels in ascending order is NOIR: nominal; ordinal; interval; and ratio. The levels of measurement describe the type of data obtained. Nominal data include categorical dataVdata that are in named (nominal) categories: yes/no, present/absent, male/female, and group assignment. Nominal data may include more than two categories such as group assignment to experimental, control, or standard-of-care group. Ordinal data also include categorical data, but unlike nominal data where the groups differ in name but not in order, ordinal data place an order to the categories. The distance between categories is unknown, but the order matters. Examples of ordinal data may include (1) socioeconomic status: categorized as low, middle, or high or (2) pain: characterized as absent, minimal, mild, moderate, or severe. Interval data are ordinal data where the distance between categories is of equal intervals and there is a true zero. Examples of interval data include temperature and pain using a 0- to 10-point scale. Finally, ratio data refer to meaningfully ordered categories with equal intervals between them and a true 0 point. Examples of ratio data include blood pressure, heart rate, and weight. Case Study Case Study The hypotheses generated by the CNS will be dependent on the literature review and supporting data. Suppose that previous studies address the concepts of spirituality, QOL, and number of ER visits in patients with HF,5Y14 but only QOL has been addressed as a DV in a study of a CNS intervention in patients with HF.10 However, spirituality, QOL, and ER visits postdischarge could serve as DVs in our case study. QOL would have data to support a hypothesis. QOL, and ER visits postdischarge. The CNS determines that the concepts will be measured using the instruments and data type as indicated below. Then, the CNS determines how many patients are needed in each group (intervention versus standard of care). 132 Concept CNS palliative care consultation intervention Measure Data Type Investigator-developed item (yes/no) Nominal www.cns-journal.com Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. May/June 2012 Spirituality Spirituality (visual analog scale 0Y10)15 Interval QOL Short-Form 1216 Ordinal ER visits postdischarge Investigator-developed item (no. of visits) Ratio In order to determine the number of patients or subjects needed in each group, a power analysis may be performed. Power is the ability to detect real differences or relationships that exist in the population from which the sample is drawn. Power is calculated using a mathematical formula based on the degree of significance, the expected effect size, and the desired level of power.17Y18 There are online calculators that compute the minimum required sample size using the P value, number of IVs, anticipated effect size (0.02, 0.15, and 0.35 and small, medium, and large, respectively), and the desired statistical power level (usually Q 0.8). One such calculator that may be useful to CNSs is the free A-priori Sample Size Calculator for Multiple Regression from Daniel Sopers Free Statistics Calculators web site available at http://danielsoper.com/statcalc3/calc.aspx?id = 1. Now that the needed sample size is known, the CNS is ready to obtain the appropriate institutional review board approvals and begin data collection. When the CNS begins data analysis, the first step is to plot the data in order to note the distribution and mode of the data. The next step is to determine whether the distribution is normal or not. Applying Statistics to Clinical Nurse Specialist Practice Applying Statistics to Clinical Nurse Specialist . A normal distribution is characterized by a unimodal, symmetrical, bell-shaped curve, when interval data are represented by a histogram or line graph.19(p128) Normal distribution may be determined by visually inspecting the histogram or line graph and analyzing the similarity of the mean and median. The closer to equal the mean and the median are, the closer to a normal distribution the data are.19 The second step is to determine the critical significance level, specifically the risk the CNS is willing to take about being incorrect. The critical significance level is represented as the alpha and is the probability that the relationship between or difference in means between groups is not happening by chance. Typically, the ! level is set at .05, which means that if P is less than .05, those differences are not due to chance. The significance level enables judgment of the evidence against the null hypothesis or evidence for the alternative hypothesis. The P value ranges from 0.00 to 1.00, with a large P value supporting the null hypothesis (there is no difference) or a small P value supporting the alternative hypothesis (there is a difference). Usually, an significance level of .05 is acceptable unless the risk of a type I error, false positive, or an incorrect decision is great. In the case of high error risk, such as a clinical drug trial, then a significance level of .01 is used. A Clinical Nurse Specialist A significance level of .05 means that one can be 95% confident in our findings or have a 5% chance of error. A significance level of .01 means one can be 99% confident in our findings, or have a 1% chance of error. Case Study After having determined the distribution and mode of the data, the CNS would need to ascertain whether the distribution is normal or not. Then, the CNS would need to decide the acceptable significance level. Next, the issue of a one-tailed or a two-tailed test must be addressed. A one-tailed test seeks either an increase or a decrease in the measured concept and is used when the research hypothesis states direction. A two-tailed test seeks any change (either an increase or a decrease) in the concept and is used when sufficient knowledge is not available to predict direction in the measured concept. Case Study In the proposed CNS case study, limited previous knowledge facilitates limited prediction of the direction of the measured concepts. Thus, a two-tailed test will be used in the statistical analysis. The next task in statistical analysis is to determine the appropriate analysis. First, the level of the research, if determined, may facilitate selection of the appropriate statistical test. There are three levels of quantitative research: descriptive, comparative/correlational, and quasi-experimental/ experimental. The research levels and the associates statistical tests are described in ascending order in Tables 1 and 2. If a level II or III study is conducted, then additional questions about the data must be considered in order to determine the appropriate statistical test. Are the data normally distributed? Are the data independent? Is the sample variance equal to the population variance? Are the data measured on an interval or ratio scale, and are Table 1. Level 1 Research Data Level Type of Research Analysis Level I Descriptive Descriptive statistics Nominal or Ordinal Interval or Ratio Frequency: number of observations in each category Mean: sum of values divided by the number of values Displayed as frequency tables, histograms, graphs, figures Median: middle value of a set of ordered numbers; mode: most frequent value or category in a distribution www.cns-journal.com Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 133 Using Research to Advance Nursing Practice Table 2. Levels II and III Research Levels Type of Research Analysis Parametric Nonparametric Level II Comparative: differences between means Level I plus inferential statistics t Test: 2 groups; interval or ratio data Mann-Whitney (U) test: 2Applying Statistics to Clinical Nurse Specialist Practice Applying Statistics to Clinical Nurse Specialist .random independent samples; at least ordinal scale (a) 2 Independent groups: independent t test Kruskal-Wallis test (K): Q2 unmatched (unrelated) groups; nominal, ordinal, or interval scale; if only 3 groups, must be Q5 observations in each group (b) 2 Matched (related) groups: paired t test Wilcoxon test (W): 2 matched (related) groups; interval scale Analysis of variance (df, F value, and P level): Q2 unmatched (unrelated) groups Correlational Level I plus inferential statistics Univariate # 2 Test: nominal scale Pearson R: an interval scale Spearman D : ordinal, interval or ratio scale Multivariate Kendall I : ordinal scale Regression: continuous variables Level III Quasi-experimental Experimental All available statistical tests and/or analyses Path analysis (causal modeling) Confirmatory factor analysis Structural regression models the data continuous? If the answer to all of these questions is yes, then parametric analyses may be used. If the answer to any of these questions is no, then nonparametric analyses must be used. With nominal or ordinal data, nonparametric statistics are commonly used. The selection of the appropriate statistical test is based the number of the groups being analyzed, and the data scale, also (Tables 1 and 2). A helpful interactive site that facilitates selection of the correct statistical test needed is available at http://statistics.laerd.com/selecting-statistical-tests/. Once the appropriate test is determined, the analyses are conducted, and the statistics are assessed for statistical significance. If P is less than the predetermined critical ! level (P G !), the probability is small that difference or relationship occurred by chance. Thus, the null hypothesis is rejected and the alternative hypothesis accepted. If P is more than critical ! level (P 9 !), the probability is high that the difference or relationship occurred by chance. Thus, the null hypothesis is accepted and the alternative hypothesis is rejected. Thus, in summary, statistics are tools in the armament of the CNS to facilitate the answering of important clinical questions. Therefore, knowledge of statistics is essential for the CNS as this knowledge enables the evaluation of research by others and guides the research work of the CNS. The findings of such research describe the data in 134 Case Study The research questions in the CNS case study seek to describe the level of the DVs. Assuming the data are normally distributed, the needed statistical analysis is described below. Concept Data Type Statistical Analyses Spirituality Interval Mean, median, mode QOL Ordinal Mean, median, mode of each subscale score ER visits postdischarge Ratio Mean, median, mode Spirituality Interval Pearson R QOL Ordinal Spearman D ER visits postdischarge Ratio Pearson R The CNS seeks to compare patients who receive the CNS consultation intervention regarding palliative care with those who receive usual care. Assuming the data are normally distributed and the two groups are independent. Concept Data Type Statistical Analyses CNS palliative care consultation intervention Nominal Frequency, percent Spirituality Interval Independent t test QOL Ordinal Mann-Whitney U ER visits postdischarge Ratio Independent t test www.cns-journal.com Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. May/June 2012 meaningful terms, contribute to the disciplines knowledge base, and augment our understanding of clinical issues. References 1. Hazard Munro B. Statistical Methods for Health Care Research, Volume 2. Philadelphia, PA: Lippincott Williams & Wilkins; 2005. 2. Keith ME. A trial of thiamin supplementation in patients with heart failure. 2011. http://clinicaltrials.gov/ct2/show/NCT00959075. Accessed February 21, 2012. 3. Albert NM, Hancock K, Murray T, et al. Cleaned, Applying Statistics to Clinical Nurse Specialist Practice Get a 10 % discount on an order above $ 100 Use the following coupon code : NURSING10

Use Promo Code: FIRST15

**FIRST15**and enjoy expert help with any task at the most affordable price.