Observational Survey
Example of an Observational (Survey) Study
Appel, S.J., Harrell, J.S., & Deng, S. (2002). Racial and socioeconomic differences in risk factors for cardiovascular disease among Southern rural women. Nursing Research, 51(3), 140-147.
The title appropriately indicates the target population sampled (Southern rural women), as well as two key independent variables (race and socioeconomic status—inclusive of education and income) examined for their associations with risk of cardiovascular disease. A possible enhancement of the study title would be to include reference to the type of study design used (e.g., “An observational study of…”). The abstract is well structured per the journal standards and is complete in presenting the key points of …show more content…
each section of the article.
The clinical problem and study objectives are clearly presented in the abstract and text of the article.
Specific research questions are outlined just prior to the Methods section of the article. A thorough and current literature review is presented in which the authors identify the current state of knowledge and needs for additional research. They make a clear and convincing case of the importance of the research for preventing cardiovascular mortality and morbidity in at-risk/vulnerable populations, of which their sample includes a subset (African American women residing in the rural southeastern U.S.). This research topic and the study population match well with current national research funding priorities focusing on health disparities in vulnerable populations (in this case, disparities among African American versus white women in the burden of mortality and morbidity from cardiovascular …show more content…
disease).
There are two limitations of the study background that if addressed could have enhanced the clarity of the presentation. First, a variety of factors that could influence the dependent variable/outcome of interest are addressed, but the logic diagram presented in Figure 1 is less than fully articulated/informative, although it presents a hypothesized pathway of how contextual risk factors ultimately affect cardiovascular risk status. The figure might be enhanced if it included specific factors that are addressed in the literature review. Second, based on the information presented in the literature review, the authors could have presented specific hypotheses to be tested about the expected relationships between the key independent variables (race, socioeconomic status—including variables of income and education) and the dependent variable (cardiovascular risk status).
An observational study design was used. This design is appropriate to address the level of research questions posed, which focused on describing associations between the independent and dependent variables of interest. Survey data were collected at one point in time. The study design is weak for making causal inferences but appropriate for exploring associations among variables that might be targets for intervention in future (intervention) studies.
The information about the sampling design is somewhat limited in detail. It appears that this study relied upon analysis of data from survey data collected from mothers participating in a research project on cardiovascular health in children and youth. The survey data were collected from 1,110 women (27% African American, 73% white) residing in several rural counties in the southeastern U.S. (North Carolina). Surveys were mailed to all mothers and fathers of children taking part in the child and youth study (an exhaustive sampling approach, in which all members of a sampling frame are sampled), and data were analyzed for the mothers from the reported-upon study who reported their race as white or African American (N = 1,110). The exhaustive sampling approach is a strength of this study. However, the response rate of parents to the survey is not reported, and it is also reported that only 35% of the eligible children participated in the survey. It is not clear if both parental and child participation (versus parent or child participation) in the study was necessary. Due to the limited/unclear response rate, the survey respondents who provided data for the reported-upon study may not necessarily have been representative of the parents in the counties from which they were sampled.
The presentation of information about the study procedures and measures is generally adequate to understand how the data were collected/measured. Table 1 presents the cardiovascular risk index (including scoring information) that was used to quantify the dependent variable in the study. This measure was developed for the purposes of the study, but little information is presented about how the measure was developed, and no information is presented about either the reliability and validity of the measure or the rationale for the scoring criteria that were used.
It appears that appropriate statistical procedures were used to address the study questions.
Significance levels (P values) for results are presented in the tables and text as appropriate. Tables 2 and 3 present the descriptive statistics for the sociodemographic variables of income and education, as well as cardiovascular diseases and other physiologic variables such as age and BMI that are associated with cardiovascular disease risk. Chi-square tests were used to test for differences in proportions of African American and white women in various nominal-level categories of income, education, and disease status (Table 1). Independent samples t-tests were used to test for group mean differences for African American and white women on physiologic variables (Table 2). Analysis of variance (ANOVA) was used to test levels of income, race, and education as predictors of cardiovascular risk. An analysis of covariance (ANCOVA) was used in which BMI was treated as a covariate, with race, education, and income treated as independent variables in predicting cardiovascular risk status. The analyses allowed the researchers to better understand how the variables of race, education, income, and BMI were associated with each other, and resulted in the conclusion that educational level and BMI were the only significant predictors of cardiovascular risk, after adjusting for income and
race.
Because the specific response rates for variables analyzed from the survey are not reported in the analyses (it seems less than plausible that there would be absolutely no missing data in analyses for a postal survey), it is not clear from the results if item non-response bias was present in the results of this study. This remains a key open question as relevant to the interpretation of the appropriateness of the statistical procedures and the validity of the results.
The Methods section mentions that the study was approved by a “multiple assurance IRB” (IRB certified as meeting Department of Health and Human Services research standards in a multiple assurance of compliance agreement). There is no specific mention of protection of the research participants’ confidentiality of responses or informed consent procedures, which were presumably incorporated in the approved IRB protocol. There are no obvious ethical problems with the reported-upon study, but the information provided is not adequate to evaluate the adequacy of protection of rights of human subjects.
The conclusions address possible implications of the findings for nursing interventions (e.g., focusing interventions on weight reduction among less well-educated women). Possible theoretical explanations for the study findings as well as directions for future research are suggested. Appropriate limitations of the study are discussed, such as limits to generalizability and causal inferences. The overall presentation is clear and consistent with conventions for scientific writing, although some details about the study methods are lacking that would be needed to do a more comprehensive critique.
Appel, S.J., Harrell, J.S., & Deng, S. (2002). Racial and socioeconomic differences in risk factors for cardiovascular disease among Southern rural women. Nursing Research, 51(3), 140-147.
The title appropriately indicates the target population sampled (Southern rural women), as well as two key independent variables (race and socioeconomic status—inclusive of education and income) examined for their associations with risk of cardiovascular disease. A possible enhancement of the study title would be to include reference to the type of study design used (e.g., “An observational study of…”). The abstract is well structured per the journal standards and is complete in presenting the key points of …show more content…
each section of the article.
The clinical problem and study objectives are clearly presented in the abstract and text of the article.
Specific research questions are outlined just prior to the Methods section of the article. A thorough and current literature review is presented in which the authors identify the current state of knowledge and needs for additional research. They make a clear and convincing case of the importance of the research for preventing cardiovascular mortality and morbidity in at-risk/vulnerable populations, of which their sample includes a subset (African American women residing in the rural southeastern U.S.). This research topic and the study population match well with current national research funding priorities focusing on health disparities in vulnerable populations (in this case, disparities among African American versus white women in the burden of mortality and morbidity from cardiovascular …show more content…
disease).
There are two limitations of the study background that if addressed could have enhanced the clarity of the presentation. First, a variety of factors that could influence the dependent variable/outcome of interest are addressed, but the logic diagram presented in Figure 1 is less than fully articulated/informative, although it presents a hypothesized pathway of how contextual risk factors ultimately affect cardiovascular risk status. The figure might be enhanced if it included specific factors that are addressed in the literature review. Second, based on the information presented in the literature review, the authors could have presented specific hypotheses to be tested about the expected relationships between the key independent variables (race, socioeconomic status—including variables of income and education) and the dependent variable (cardiovascular risk status).
An observational study design was used. This design is appropriate to address the level of research questions posed, which focused on describing associations between the independent and dependent variables of interest. Survey data were collected at one point in time. The study design is weak for making causal inferences but appropriate for exploring associations among variables that might be targets for intervention in future (intervention) studies.
The information about the sampling design is somewhat limited in detail. It appears that this study relied upon analysis of data from survey data collected from mothers participating in a research project on cardiovascular health in children and youth. The survey data were collected from 1,110 women (27% African American, 73% white) residing in several rural counties in the southeastern U.S. (North Carolina). Surveys were mailed to all mothers and fathers of children taking part in the child and youth study (an exhaustive sampling approach, in which all members of a sampling frame are sampled), and data were analyzed for the mothers from the reported-upon study who reported their race as white or African American (N = 1,110). The exhaustive sampling approach is a strength of this study. However, the response rate of parents to the survey is not reported, and it is also reported that only 35% of the eligible children participated in the survey. It is not clear if both parental and child participation (versus parent or child participation) in the study was necessary. Due to the limited/unclear response rate, the survey respondents who provided data for the reported-upon study may not necessarily have been representative of the parents in the counties from which they were sampled.
The presentation of information about the study procedures and measures is generally adequate to understand how the data were collected/measured. Table 1 presents the cardiovascular risk index (including scoring information) that was used to quantify the dependent variable in the study. This measure was developed for the purposes of the study, but little information is presented about how the measure was developed, and no information is presented about either the reliability and validity of the measure or the rationale for the scoring criteria that were used.
It appears that appropriate statistical procedures were used to address the study questions.
Significance levels (P values) for results are presented in the tables and text as appropriate. Tables 2 and 3 present the descriptive statistics for the sociodemographic variables of income and education, as well as cardiovascular diseases and other physiologic variables such as age and BMI that are associated with cardiovascular disease risk. Chi-square tests were used to test for differences in proportions of African American and white women in various nominal-level categories of income, education, and disease status (Table 1). Independent samples t-tests were used to test for group mean differences for African American and white women on physiologic variables (Table 2). Analysis of variance (ANOVA) was used to test levels of income, race, and education as predictors of cardiovascular risk. An analysis of covariance (ANCOVA) was used in which BMI was treated as a covariate, with race, education, and income treated as independent variables in predicting cardiovascular risk status. The analyses allowed the researchers to better understand how the variables of race, education, income, and BMI were associated with each other, and resulted in the conclusion that educational level and BMI were the only significant predictors of cardiovascular risk, after adjusting for income and
race.
Because the specific response rates for variables analyzed from the survey are not reported in the analyses (it seems less than plausible that there would be absolutely no missing data in analyses for a postal survey), it is not clear from the results if item non-response bias was present in the results of this study. This remains a key open question as relevant to the interpretation of the appropriateness of the statistical procedures and the validity of the results.
The Methods section mentions that the study was approved by a “multiple assurance IRB” (IRB certified as meeting Department of Health and Human Services research standards in a multiple assurance of compliance agreement). There is no specific mention of protection of the research participants’ confidentiality of responses or informed consent procedures, which were presumably incorporated in the approved IRB protocol. There are no obvious ethical problems with the reported-upon study, but the information provided is not adequate to evaluate the adequacy of protection of rights of human subjects.
The conclusions address possible implications of the findings for nursing interventions (e.g., focusing interventions on weight reduction among less well-educated women). Possible theoretical explanations for the study findings as well as directions for future research are suggested. Appropriate limitations of the study are discussed, such as limits to generalizability and causal inferences. The overall presentation is clear and consistent with conventions for scientific writing, although some details about the study methods are lacking that would be needed to do a more comprehensive critique.