The first article that will be examined in this paper is Impact of Hospital Nursing Care on 30-Day Mortality for Acute Medical Patients by Tourangeau1, A.E., Doran, D.M., McGillis Hall, L., O’Brien Pallas, L. and Pringle, D.
Review of Data Analysis
Brief Statistic Summary
The study represents two regression models: multiple regression model and backward regression model. Some of the assumptions were met while some of them were not. The type of data can be considered appropriate for these statistical tests because the results of the tests supported the assumptions that were made at the beginning of the study (Bellolio, Serrano and Stead, 2008).
Research Problem
The research problem is to determine and examine the nursing-related determinants of risk-adjusted 30-day mortality for acute patients in Canada (Ontario).
Brief Overview of Data Collecting Approach
Data was collected with the help of analysis of primary and secondary data. Secondary data was collected with the help of four databases, namely: the Ontario Discharge Abstract Database 2002–2003 (DAD) was used for selection of patients their health-related information; Statistics Canada 2001 Population files was used for analysis of the patients according to socio-economic indicators; the Ontario Hospital Insurance Plan database for development of proxy indicators for patients’ preadmission health status; the Ontario Registered Persons Database for identification death dates within 30-days period after admission to a hospital. The data collected from the DAD, the Ontario Hospital Reporting System file 2002–2003 (OHRS), and the Ontario Nurse Survey 2003 was used to create variables. The data collected from the Ontario Nurse Survey 2003 was the only primary source (Tourangeau1, Doran, McGillis Hall, O’Brien Pallas, Pringle, Tu and Cranley, 2006).
Sources of Data
Data for analysis was derived from existing statistical information provided by Statistics Canada
2001 Population files and other sources of information mentioned above. Ontario Nurse Survey 2003 provided survey data regarding 30-day mortality for acute patients in Ontario. DAD database provided demographic data and the codes for responsible diagnosis and other related information (Tourangeau1, Doran, McGillis Hall, O’Brien Pallas, Pringle, Tu and Cranley, 2006).
Independent and Dependent Variables
Independent variables are nursing-related determinants. There were 19 independent variables identified for predictors of mortality, namely: nursing staff mix, nursing staff dose, proportion of full-time nursing staff, years nurse experience on current clinical unit, proportion of baccalaureate-prepared nurses, overall nurse level of health, missed work hours, nurse–physician relationships, manager ability and support, adequacy of staffing and other resources, teamwork in care delivery, overall nurse job satisfaction, nurse reported quality of care on unit, nurse burnout, amount of professional role support, use of care maps/protocols to guide patient care, physician expertise, outside urban area hospital, urban non-teaching hospital, and teaching hospital (left out of models as reference group). Dependent variables are occasions of mortality within 30-days from the date of admission to a hospital.
Estimation of Sample Size
There were several hypotheses developed, some of them were supported, some of them were partially supported, some of them were rejected. There was large sample size of 46,993 patients was used. The sample size of nurses surveyed made up 5,980. Power analysis was conducted because sample sizes are quite large (Tourangeau1, Doran, McGillis Hall, O’Brien Pallas, Pringle, Tu and Cranley, 2006).
Statistical Power Evaluation
The sample size was not appropriate for the statistical test that was implemented. Least squares regression analysis was implemented to meet the research assumptions. This method allows entering all variables simultaneously. The sample size was not appropriate because statistical power is lower than 0.8 that cannot be considered normal (Eng, 2003).
Measurement of Hypotheses
There were several hypotheses were analyzed within the study. The statistics used in the research were appropriate. However, they were taken from previous research. The study insignificantly contributed to the existing research. The assumptions were met on the basement of two-model regressions. The first regression model used regression coefficients, t-statistics, p-values, and standard errors. These statistics showed correlation between lower mortality in the hospitals and its predictors sorted out from the previous research (Walker, 2005). Only fine of nineteen predictors were found significant. Other variables did not add significantly to the research. F-value and r2-value used in the backward regression model showed that eight predictors added to identification of mortality rates (Plichta and Kelvin, 2013).
Discussion of Displaying Data
Data were displayed with the help of the tables. Data in the tables were well-organized and clear because any information needed can be easily found in the tables. However, using diagrams and graphs would help better visualize the results of the research (Walker, 2005).
Data Analysis Evaluation
Findings Analysis
Data analysis showed the dependency of mortality and education of nurses, namely: nursing staff dose and proportion of Registered Nurses in hospital staff mix. There was an inverse relationship also found meaning preferences in resource allocation for nursing services. Registered Nurses cost more for a hospital. However, it was found that 10% increase in Registered Nurses staffing is associated with 6 fewer deaths per thousand of discharged patients based on regression coefficient. Similar findings were found when analyzing the correlation between dose of nurse staffing and patients’ deaths. Another important finding was that hospitals having higher proportion of baccalaureate-prepared nurses tend to have lower mortality rates within 30 days after hospitalization of acute medical patients. Thus, 10% increase in number of baccalaureate-prepared nurses in the hospital is bind to nine fewer deaths per a thousand of patients. Also, positive correlation was found between nurse-reported staffing adequacy and resources. Thus, 10% increase in staffing adequacy is associated with 17 fewer deaths per a thousand of patients (Tourangeau1, Doran, McGillis Hall, O’Brien Pallas, Pringle, Tu and Cranley, 2006).
The most important finding was that positive correlation between using care maps and lowering mortality was found. Besides, lower mortality rates were found in the hospitals having nurses with average emotional exhaustion at their working place.
Conclusion Analysis
In the course of research the following conclusions were drawn: the hospitals should maximize the number of Registered Nurses providing care for acute medical patients, increase the baccalaureate-prepared nurses to provide care for acute medical patients, and making investments in updating care maps guiding patient progress during hospitalization. The most important conclusion was that only 45% of variance was explained within the developed statistical model while 55% of the variance was not explained.
Study limitations were discussed in the article. There were several limitations outlined, namely: the sample of 75 hospitals was not appropriate conditioning weak power of regression models; second limitation relates validity of internal and external studies. The factors that could be significant for the research result were not included into the models that weakened internal validity of the study. The impact of those factors was not taken into account in the study. In addition, new determinants of acute medical patients’ mortality within 30-days period after hospitalization should be developed because the determinants that were developed in this study did not reflect the reasons of patients’ deaths at fullest because not all factors of mortality were considered (Tourangeau1, Doran, McGillis Hall, O’Brien Pallas, Pringle, Tu and Cranley, 2006).
Two regression models were well-developed. The models could be used for further research. However, the models could be expanded when new mortality determinants will be revealed. The lack of mortality determinants conditioned high rate of variance that was not explained by the researchers. The results can be trusted because they support logical conclusions that can be made in the process of the research.
Data Interpretation
The case study supports the hypotheses that were developed. However, most of the hypotheses were developed within the framework of the previous research; thus, the study does not contribute much to the research of the rate mortality of acute medical patients.
The most confusing finding was that only 45% of variance was explained within the study, while the other half remained unexplained due to the wrong hypotheses made. The main aspect of the study that must be questioned is that to find other determinants influencing mortality rate of acute medical patients within 30 days after admission to hospitals. The study provides sufficient evidence to change practices because the results of the research support the natural course of events. For example, correlations between nurses’ appropriate education and the rate of mortality.
The statistics used in the current research are typically used for the research conducted in medicine. The statistics can be considered useful because it supported the majority of the hypotheses that were developed. The problem of the research was not in using statistics, but in developing the right hypotheses (SOCR, 2002).
Analysis of the Second Article
The first article that will be examined in this paper is Prevalence of Obesity among Nigeria Nurses: the Akwa Ibom State Experience by Ogunjimi1, L.O., Ikorok, M.M. and Olayinka, Y.O.
Review of Data Analysis
Brief Statistic Summary
There was a phenomenon of obesity studied on the example of nurses working in Akwa Ibom State Public Health Institutions. Data were analyzed with the help of percentage frequency. Type of data collected is appropriate for conducting statistical test used in the research. The assumptions of the study were met. Except for percentages frequency, Pearson’s product moment correlation, population t-test, and chi square test were implemented (Plichta and Kelvin, 2013).
Research Problem
The primary concern of the research conducted by Ogunjimi1, Ikorok and Olayinka (2010) is weight of health personal and examination of obesity prevalence in Nigeria.
Brief Overview of Data Collecting Approach
The sample used in the research consists of 500 nurses that were selected from three zones of health in Akwa Ibom State. There were 200 nurses selected from Uyo zone; 150 nurses were selected from Eket and Ikot-Ekpene zones. The higher percentage of nurses selected from Uyo zone is conditioned by the fact that more nurses is presented in this zone. The populations examined in the course of the research have government trained female nurses mainly. The population is almost a half of the total number of nurses working in public health institutions in Akwa Ibom State (Ogunjimi1, Ikorok and Olayinka, 2010).
Sources of Data
There were two data sources used in the research, namely: a structured questionnaire and BMI table. The questionnaire consisted of two sections containing personal data and the second part containing 16 items likert-scale. Personal data collected was used for analysis of bodily mass index and marital status. The second section was used to identify eating habits and job schedule. For the purposes of determining reliability Cronbach’s coefficient was used. In order to calculate the level of obesity, data derived from World Health Organization was analyzed. The questionnaire was examined by two professionals in psychology. Further, questionnaire was certified for validity and used in this study (Ogunjimi1, Ikorok and Olayinka, 2010).
Independent and Dependent Variables
There was one dependent variable and two independent variables used in the current research. A dependent variable in the case that was examined was the level of obesity of the nurses while eating habits and marital status are independent variable used in the research. Thus, the main assumption was that obesity was associated with eating habits and marital status of the populations of the nurses that was examined (Ogunjimi1, Ikorok and Olayinka, 2010).
Estimation of Sample Size
The size of the sample allows suggesting that power analysis was conducted because it is quite large for the statistical tests used in the analysis. The total number of nurses surveyed was 500. Power analysis was conducted because this sample size meets the requirements of the statistical tests that were implemented in this research (Eng, 2003).
Statistical Power Evaluation
The level of significance was 0.05 and the critical t-values equaled 1.95 for one-tailed test. Thus, the results of the study are statistically significant. For the current statistical analysis the following formula should be used to identify if statistical power analysis was conducted:
The sample size should not be less than that providing statistical power not lower than 0.8. In this case statistical power analysis was conducted because the sample size was large enough for the purposes of the current research (Eng, 2003). Slightly less than a half of the nurses, working in Akwa Ibom State were surveyed (500 out of 1082), showing more than 62% of obese persons. Thus, the results of the research can be considered relevant (Ogunjimi1, Ikorok and Olayinka, 2010).
Measurement of Hypotheses
The hypotheses were not clearly outlined in the study. However, they can be derived from the results of study analysis. The assumptions were met through conducting several statistical tests. Thus, percentage of obesity frequency was calculated to determine how often the cases of obesity occur in the sample. Population t-tests analysis was conducted to identify average BMI in the sample and compare it to the normal BMI in order to reveal whether obesity among Nigerian nurses took place. Pearson’s correlation analysis was conducted to identify whether eating habits and marital status influence obesity index. Contingency Chi-square analysis was conducted to identify correlation between marital status and obesity index (Plichta and Kelvin, 2013).
Discussion of Displaying Data
Data were displayed with the help of the tables with the comments regarding statistical significance, critical values, and sample size beneath each table related statistical test. Data were visualized appropriately so the results of each test can be seen (Walker, 2005).
Data Analysis Evaluation
Findings Analysis
Data analysis showed the dependency between obesity index, eating habits, and marital status of the participants. The following conclusions were drawn during the research process: the majority of the participants (62.6%) suffered from obesity according to the research results; eating habits played the most significant role in the level of obesity; married nurses appeared to be more obese than their unmarried or divorced peers.
The conclusions drawn were appropriate since they were supported by statistical evidence. Besides, the results of the study were supported by the results of other studies. Thus, the findings made by Ogunjimi1, Ikorok and Olayinka (2010) intersect with the findings developed by Diehl (1990) and Labib (2004). Comparison of the positive r-value of 0.45 and critical r-value 0.198 with 0.05 level of significance and 498 degrees of freedom indicated the connection between indiscriminate eating and obesity. The influence of marital status on obesity can be questioned because of difficulty to compare the results that were developed using different statistical methods. However, the studies of Lipowk-z et al. (1998) and Sobal et al. (2002) who reported correlation between marital status and obesity, as well as the current study. Besides, Lee et al. (1999) reported decrease in BMI between women who were divorced and those who were unmarried (P < 0.01). Thus, the results of the research proved to be relevant with regard to the factors that were identified. The limitations of the study were not discussed (Ogunjimi1, Ikorok and Olayinka, 2010).
Conclusion Analysis 150 +40
The sample was collected carefully and thoughtfully. The questionnaires were examined by psychologists thus increasing the level of relevance. Using stratified proportionate sampling technique was statistically relevant because it allowed grouping the sample according to three health zones. Utilization of 16 likert-scale allows receiving detailed information regarding the research questions.
More factors influencing obesity should be included in the research. The analysis of only two determinants of nurses’ obesity does not allow sufficient evidence of obesity reasons. There should be psychological aspect of obesity studied as well. In addition, such factors as social pressure, culture, and life style should be also included in the research in the future.
The findings developed in the study shed the light on correlation between eating habits, marital status, and obesity. The results can be trusted because the statistical tests used in the research offered high level of significance. However, the current findings could be expanded at the expense of including greater number of obesity determinants.
Data Interpretation
The study was useful because it helped reveal the factors influencing obesity of Nigerian nurses. The authors proposed recommendations based on the results of the statistical tests that showed correlation between two determinants and the level of obesity.
The study can be expanded with more determinants and more advanced statistical methods helping analyze the problem in details. The correlation between obesity and marital status was found confusing because peers did not consider this factor significant for the level of obesity. Another point of concern was that some of other authors, such as Galassie (2004) and Aldair 2005), reported quite different results that can be conditioned by using alternative methods of statistical research. This aspect of the study can be questioned. Besides, the reasons of obesity were not studied in details because the researches included only two obesity determinants. The study provides sufficient evidence to change practice and relevant recommendations regarding further actions aiming to improve the situation.
The statistics used in the research are typical to those used in medical research. Thus, they can be considered useful because they are widely used in medicine.
References
Bellolio, M.F., Serrano, L.A. and Stead, L.G. (2008). Understanding statistical tests in the medical literature: which test should I use? International Journal of Emergency Medicine, 1(3), 197–199. doi: 10.1007/s12245-008-0061-z. Retrieved from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2657277/
Radiological Society of North America, 227, 309-313. doi: 10.1148/radiol.2272012051.
Retrieved from http://radiology.rsna.org/content/227/2/309.full
Ogunjimi1, L.O., Ikorok, M.M. and Olayinka, Y.O. (2010). Prevalence of obesity among Nigeria nurses: the Akwa Ibom State experience. International NGO Journal, 5(2). Retrieved from
http://www.academicjournals.org/ingoj/pdf/pdf2010/Feb/Ogunjimi%20et%20al..pdf
Plichta, S.B. and Kelvin, E. (2013). Munro's statistical methods for health care research. 6th ed. New York: Wolters Kluver Health.
SOCR Resource. (2002, Jan.1). Choosing the right test. SOCR, Retrieved from
http://www.socr.ucla.edu/Applets.dir/ChoiceOfTest.html
Tourangeau1, A.E., Doran, D.M., McGillis Hall, L., O’Brien Pallas, L., Pringle, D., Tu, J.V. and Cranley, L.A. (2006). Impact of hospital nursing care on 30-day mortality for acute medical patients. Journal of Advanced Nursing, 57(1), 32–44. doi: 10.1111/j.1365-2648.2006.04084.x
Retrieved from
http://www.researchgate.net/publication/6616822_Impact_of_hospital_nursing_care_on_30-day_mortality_for_acute_medical_patients/file/9fcfd50af00d9cad16.pdf
Walker, W. (2005). The strengths and weaknesses of research designs involving quantitative measures. Journal of Research in Nursing, 10(5). 571-582. DOI: 10.1177/136140960501000505.
Retrieved from http://www.sagepub.com/gray/Website%20material/Journals/jrn_walker.pdf