Ensayos sin conclusiones: la necesidad de preparación del subsuelo
3 I EL PROYECTO GENERAL DE URBANIZACIÓN DEL ENSANCHE
The digitally recorded interviews were transcribed verbatim by a professional transcriptionist, and verified accuracy PI by comparing the transcription to the recorded interview (Sandelowski, 1995). The transcribed interviews, field notes, and other text data were stored in an electronic database. The team used ATLAS.ti qualitative data analysis software system (Silver & Lewins, 2009) to aid in coding, organizing, and managing the data. Each document was labeled by source (family member, patient) and by type (interview, observation).
4.3.2 Data Analysis
4.3.2.1 First level analysis
Analysis began with a reading of each document as it became available from the transcriptionist. I read each document several times to gain an understanding of the whole context (Graneheim & Lundman, 2004; Sandelowski, 1995). During this initial read, I conducted manifest coding and also noted any latent concepts that became apparent (Sandelowski, 1995). To increase the trustworthiness of findings, throughout the entire coding process I used memos to keep an audit trail of any coding decisions and development of ideas about explanations and themes (Sandelowski, 1995). In addition, I used reflexive memos to explore assumptions (Hsieh & Shannon, 2005).
4.3.2.1.1 Provisional coding
To answer the first research question, “What are the specific strategies that nurses use when supporting family members making EOL decisions in the ICU during trajectory of the decision-making process from the perspective of the family members?,” I used a deductive method called provisional coding (Saldana, 2009). Provisional codes are codes that have been determined prior to entering the field, and are based on literature reviews, conceptual frameworks, the researcher’s own previous experience, and results of pilot studies (Saldana, 2009). Although provisional codes are based on a priori concepts, they are not necessarily fully developed definitive concepts (Blumer, 1954). The term provisional indicates that during the coding process, changes might have been made to the definitions and understanding of these concepts (Saldana, 2009). Through my review of the literature (Adams et al., 2011) and my pilot study (Adams, Bailey, Galanos, Zomorodi, & Anderson, 2010), I was able identify three roles that nurses play: information broker, supporter, and advocate. Each of these concepts has
corresponding attributes. From this data, I developed a preliminary codebook of these attributes with definitions, exclusions, and exemplars to guide my coding. See
Appendix F for a sample of this codebook. Using these a priori codes as a provisional coding framework, I coded in a deductive manner identifying and tagging units of data
that demonstrated and described these nursing roles and strategies as perceived by the participants. During this phase, I remained open to new nursing strategies that might not have been in the literature or noted in my pilot study. In addition, I was open to redefining the codes. Thus, there are many a posteriori codes and operational definitions that emerged from this phase of coding.
4.3.2.1.2 Open coding
Because the literature is unclear about how family members perceive and respond to nursing strategies, to fully answer research question one, “What are the specific strategies that nurses use when supporting family members making EOL decisions in the ICU during trajectory of the decision-making process from the
perspective of the family members?,” and to answer research question two, “How are these strategies perceived by the family members as helpful or not helpful in supporting their ability to make decisions on behalf of the patient?,” I used a more inductive
approach called open or initial coding (Bernard & Ryan, 2010). I began by examining the data in clusters to which meaningful concepts (codes) that were identified were attached (Graneheim & Lundman, 2004). The interview guide determined the
boundaries of each cluster of data. Each question and its corresponding answer defined a cluster. When a participant changed the course of the answer, and it was obvious that the content was very different, I made this a new cluster. These codes took the form of
process codes, as I identified nursing actions and family members’ responses to those actions (Saldana, 2009). The nursing strategies and roles that I identified in my previous work, as well as communication strategies identified in the medical literature, were used as sensitizing concepts to guide the search for empirical evidence of whether and how family members perceive these strategies (Blumer, 1954; Bowen, 2006).
4.3.2.2 Second level analysis 4.3.2.2.1 Pattern coding.
Once the data were coded using both a priori (structural) and a posteriori (open) codes, pattern codes were used to categorize the data and explore emerging themes (Saldana, 2009). The coded text were arranged into categories and subcategories based on how the codes were related (Hsieh & Shannon, 2005). Similar codes were placed together, examined, and given a new code to represent a category of these similar codes (Saldana, 2009). These categories and subcategories represent the manifest content of the data and were exhaustive and mutually exclusive (Graneheim & Lundman, 2004). In order to understand the latent content, or the underlying meaning of the data, categories were linked to emerging themes, which involved interpretation and explanation
4.3.2.2.2 Longitudinal coding.
The third research question, “How do these strategies change over this trajectory of decision-making from the perspective of the family members?” was answered using longitudinal coding (Saldana, 2009). This type of coding is used when a researcher wants to examine how a phenomenon changes over time and often involves constructing a matrix to compare data over time (Saldana, 2009). I examined the data over time by creating a trajectory line (Clipp et al., 1992) of the ICU stay.
I created trajectory lines for each case for which I conducted at least three interviews. Each of these trajectory lines represented the family members’ responses to the changes in the patient’s condition and the behaviors of the health care providers in a patient determined to be at high risk of dying in the ICU. Each line included the
frequency of helpful strategies identified by the family member, the condition of the patient, significant events, and numerical measures of hope for survival from the family members, physicians, and nurses caring for the patient. These trajectory lines were examined to interpret how the family members’ perception of nursing strategies changed over the trajectory of the time the participants were enrolled in the study.
The combination of these inductive and deductive coding methods allowed me to expand on the existing knowledge to provide a greater depth of understanding of how family members perceive nursing strategies in the transition from curative to palliative care in the ICU.