Types of Variables and Data Analysis

Answer

Alvarez, C., Greene, J., Hibbard, J., & Overton, V. (2016). The role of primary care providers in patient activation and engagement in self-management: a cross-sectional analysis. BMC Health Services Research, 16, 85. http://doi.org/10.1186/s12913-016-1328-3

Article critique

What research question did the authors set out to answer?

Increased emphasis on chronic care management worldwide has signaled the need for all the stakeholders in the health care system to transform from hierarchical patriarchal model, to partnerships. The partnership model recognizes the central role played by patients in self-care, for the achievement of optimal health outcomes. Primary Care Providers are seen to play a critical role in helping the patients get more involved in improving self-management and in supporting patient behavior change. Based on this background, this study set out to answer the following questions: 

• What is the relationship between Clinician Support for Patient Activation Measure (CS-PAM) and clinician behaviors that support patients’ self-management?
• What is the relationship between CS-PAM and clinician behavior in support of patients’ behavior change?
• What is the relationship between CS-PAM and actual changes in Patient Activation Measure score (PAM) among a PCPs patient panel

List the variables examined in this study.

Independent variable: Clinician support for patients’ activation CS-PAM

Dependent variables:

1. Clinicians’ engagement in chronic illness management support behaviors
2. Using specific strategies to support patient behavior change
3. Average amount of change in patient activation score a PCPs’ patient panel had over 1 year.

What type of bias is inherent to the study design used in the study?

A Crossectional is vulnerable to information bias, selection bias and self-selection bias (nonresponse) (Abebe, et. al., 2015; Pandis, 2014; Spencer, et al., 2014). Selection bias happens when participants chosen for a research happen to be different from other people not included although are eligible. In this study, participants were from only one health care organization in Minnesota and may possess characteristics that eligible participants from other states may not have, leading to biased results. The participants may also have been different in terms of the variables measured. This study registered a response rate of 68%. The non-responses (self-selection because it is committed by the participants) may have been more from the PCPs who had low views regarding patients’ activation and self-engagement. The results may thus be biased due to nonresponse bias. This design is also prone to information bias where there occur systematic errors in measurements of the variables, and is related to the use of surveys (Pandis, 2014). For example, the participants may have given socially acceptable responses rather than the actual position. Also responses may have been more accurate among participants with the desired view about patient activation and involvement, leading to differential classification.

Which type of PCP is most likely to indicate more positive beliefs about the importance of patient knowledge and involvement in his/her care? (Table 1)

Physician assistant

Identify the results that are statistically significant, list the highest tercile percentages for those items and their respective p-value. What does the difference in p-value mean? (Table 2)

• PCPs make sure the patients is involved in setting the agenda for the visit, 46.5% of respondents in the highest tercile, p<0.001
• PCPs check on progress patient is making towards behavioral goals, 8% of respondents in the highest tercile, p<0.001
• PCPs ask the patients about their personal preferences about treatment options, 44.8% respondents in the highest tercile, p<0.001
• PCPs actively involve the patient in problem solving and planning for how they will manage their health in daily life, 55.9% of respondents in the highest tercile, p<0.001
• PCPs talked to new patients about what they can expect from you as their clinician, 513% of respondents in the highest tercile, 51.3% of respondents in the highest tercile, p<0.05

The differences in the p-value correspond to the differences in the level of confidence at which to reject the null hypothesis, chosen for the particular test (Wagner, 2016). At p <0.05, the level of significance chosen is 95% and signifies a less than 1 in 20 chance of being wrong. At p < 0.01, the level of significance chosen is 99% and signifies a less than 1 in 100 chance of being wrong. At p<0.001, the level of significance chosen is 99.9% and signifies a less than 1 in 1000 chance of being wrong.

What limitations were noted with regard to the study’s findings and conclusions?

• Assessment of the PCPs behavior was based on self-reports. Self-reports do not always reflect actual behavior.
• Because the findings were based on a Crossectional design, it is not easy to determine temporal relationships between the variables studies. The analyses assumed that the views of the PCPs preceded their behaviors. From this design one cannot tell which occurred first.
• Patient panel changes in level of patient activation were derived from a different source, and were only available for a portion of patients and a portion of the clinicians. This small sample size could have limited the findings
• Data for this study were drawn from a single healthcare organization, limiting the generalizability of the findings.