University of Arkansas System
Type of paper: Thesis/Dissertation Chapter
Misclassification bias arising from random error in exposure measurement: implications for dual measurement strategies
This research study was done by Chlan and Savik, (2011) in an attempt to explore the anxiety patterns in the ICU patients using mechanical ventilation support. The research employed descriptive statistics for ordinal and interval data that were presented as medians with ranges and provided the skewed data distributions. The categorical data was analyzed and presented as frequencies. The initial analysis graphed the anxiety trajectories for every participant to discern the pattern changes. The mixed model effects were then employed in the analysis since they accommodate nonhomogeneous and correlated residuals that were expected in the repeated measures. The research underscores that the mixed models provides ideal models for analysis of data with disparate time assessment missing points of data or both from the subjects being unwilling or unable to complete the daily assessment of anxiety due to mental status, medical condition of level of fatigue. The research estimated a series of models do determine the preferred change model for the study VAS-A (Chlan & Savik, 2011).The unconditional model means were then estimated to determine the appropriateness of further modeling. Each outcome Yii combined the individual deviations and the linear of the grand mean from the grand mean. The unconditional model means were used to assess the two null hypotheses (a) no changes across occasions (b) no variation between participants. Further, an unconditional model of growth with DAY was added to predict the estimation of change coefficients (Chlan and Savik, 2011). The models with multiple within-person error covariance compatible structures with the pattern of correlation between VAS-A scores at dissimilar points of time were then explored.
The researchers employed subjects that included subgroup of participants enrolled in a multi-site, ICU-based randomized trial testing patients undergoing through mechanical ventilatory support (Chlan & Savik, 2011). The study participants were recruited from five medical centers multi-site trial representing 12 separate ICUs. The patients that were receiving mechanical ventilatory support for primary pulmonary problem such as respiratory distress who were alert and interacted with the medical staff were also enrolled to participate in the study. The study employed descriptive design and the subjects in the secondary analysis were those randomized to usual care control condition. Bordens and Abbott, (2014) writes that usual care includes the standardized nursing care protocols and standing medical orders for ever representative ICU whereby registered nurses provide care in 1:2 nurse to patient ratio. The use of randomized trials could possibly cause bias in section of the participants (Friedman, (2004). Delgado-Rodriguez and Llorca, (2004) also highlights that the use of randomized descriptive study design leads to under-representation or over-representation leading to elements of biases. The participants were enrolled at separate times during their stay in the ICU and on course of the mechanical ventilatory assistance. Therefore, there was a possibility of selection bias as a result of random sampling could be controlled by use of population-based controls or controls with disease not related to the exposure (Greenwood & Levin, 2007).
Bias and Variable Control
The number of missing scores on the VAS-A scale due to systemic error bias when the patients were fatigued to complete the assessment provided a challenge to the study. However, the study did not attempt to discern the anxiety sources and only used the anxiety ratings recorded on one assessment time point per day. While the participants were enrolled at separate times, the results of the study provide that there was no relationship between the initial ratings of anxiety obtained and the number of days in the mechanical ventilatory support and this possibly minimized the chances of selection bias in the study (Koplan, Thacker & Lezin, 1999). The dependent variable of the study was anxiety while dose frequency, sedative exposure, and time represented the independent variables. The dose frequency variable was used as a control variable to control the effects of sedative exposures. Sedative exposures to the ICU patients were instrumental since the patients received robust sedative and analgesic medications that could influence the ratings of their anxiety (Brenner & Blettner, 1993).
Bordens, K. S., & Abbott, B. B. (2014). Research design and methods: A process approach (9th ed.). New York, NY: McGraw-Hill.
Friedman, G. D. (2004). Primer of epidemiology. New York, NY: McGraw-Hill Medical. ISBN: 9780071402583.
Brenner, H., & Blettner, M. (1993). Misclassification bias arising from random error in exposure measurement: implications for dual measurement strategies. Am J Epidemiol.;138:453–461.
Chlan, L., & Savik, K. (January 01, 2011). Patterns of anxiety in critically ill patients receiving mechanical ventilatory support. Nursing Research, 60, 3.Delgado-Rodriguez, M., & Llorca, J. (2004). Bias. Journal of Epidemiology and Community Health, 58(8), 635–641.
Greenwood, D. J., & Levin, M. (2007). Introduction to action research: Social research for social change (2nd ed.). Thousand Oaks, CA: Sage Publications. ISBN: 9781412925976.
Koplan, J. P., Thacker, S. B., & Lezin, N. A. (1999). Epidemiology in the 21st century: Calculation, communication, and intervention. American Journal of Public Health, 89(8), 1153–1155.