A purposive sampling method was used at the level of selecting the hospital and units from which participants were recruited. The units were purposively selected to ensure that nurses who had experience with care for critically ill patients who commonly suffer acute pain participated in the study. All nurses at the selected units were eligible to participate in the study. The researcher approached each of the nurses available during the time of the study at the selected units and explained the purpose of the study and requested for their participat ion. Then the nurse who wished to participate would read and sign a consent form.
3.7.0 Study variables/ measurements 3.7.1 Independent variables
Demographic characteristics; age, rank, gender and level of education qualification, years of working experience, years of unit experience, pain in-service education. 3.7.2 Dependent Variables
practices related to pain assessment
Level of knowledge related to pain assessment
Perceived barriers to pain assessment and management 3.8.0 Description of study tool
Quantitative data was collected using a semi- structured questionnaire (see Appendix I). It was an instrument developed, piloted in five ICUs and re-evaluated by ten experts in pain, critical care and research methodology in Canada (Rose et al, 2011). An iterative process was
used to assess the face and content validity of the tool after reviewing existing studies on pain assessment tools, behaviors and procedures considered potentially painful. The reliability of the tool had not been reported but the experts rated the instrument’s clarity, content validity and comprehensiveness based on the method described by Burns and colleagues (2008). The tool had been used in Canada and found to be reliable (Rose et al, 2011). The researcher got permission to make modifications in the questionnaire to be relevant to the setting (see Appendix II). Specifically, the researcher modified the tool by changing the responses to some of the closed questions on practice and knowledge from the likert style (for example not at all, minimally, somewhat, moderately and extremely) to Yes or No format. Rearrangement of questions and dividing some into two parts was also done. The division of some questions was done in such a way that the participant needed to select either no or yes to continue or not with the second part .Questions about patients who unable to self-report pain were removed as the researcher selected to first concentrate on pain assessment for patients who can report pain. A few open and closed questions were added. Since the tool was used in a setting different from African type and with the modifications made, the researcher established the reliability and validity of the questionnaire. Internal consistency reliability was assessed by obtaining coefficient alpha (Cronbach’s alpha). It was used as the index to estimate the extent to which different subparts of the instrument were reliably measuring knowledge about pain assessment principles for critically ill patients. Internal consistency reliability was chosen because the measure of knowledge levels involved summing item scores (Polit & Beck, 2008). In general values, below .60 are considered poor, .60 to .69 are minimally acceptable, .70 to .79 are moderate or acceptable, .80 to .89 are considered good, above .90 are excellent. The reliability test in the study revealed the scale to be acceptable with an overall Cronbach’s alpha of the 0.71. The value reflects acceptable internal consistency, and data collected using the instrument is considered to be reliable data. Content validity was used. Content validity is the extent to which an instrument has an appropriate sample of items for the construct being measured and adequately covers the construct domain (Polit & Beck, 2008). The researcher gave the tool to 3 experts in critical care and requested them to evaluate whether individual items were relevant and appropriate. They were asked to rate the items on a 4-point scale of
relevance in relation to knowledge and practices of pain assessment principles. Where 1= not relevant, 2= somewhat relevant, 3= quite relevant &4= highly relevant. Then, for each item, the item content validity index (I-CVI) was computed as the proportion of experts in agreement about relevance by giving rating of 3 or 4. Then, the scale –level content validity (S-CVI) was computed by averaging the I-CVIs. A value of .90 was recommended as the standard for establishing excellent content validity (Polit & Beck, 2008). Discussions with some of the evaluators were done to understand the reason for the ratings and how to improve. The Questionnaire consisted of yes/ no and a few open- ended questions. The questions were arranged in four major sections namely; sections on knowledge, practices, barriers, and
demographic characteristics. The questions on knowledge were intended to yield responses that answered study question two. Some of the concepts in these questions included
knowledge on who accurately assesses patients’ pain; need to assess pain among others. The questions on practices were intended to collect responses that answered study question one. The concepts covered included whether nurses assess for pain, use of pain assessment tools, and documentation of pain scores among others. Questions aimed at answering study question three covered barriers to adequate pain assessment related to the patient, organization and provider.
3.9.0 Ethical Consideration
Ethical clearance was sought and granted from the Research Ethical committee at MUHAS and Mulago Hospital Research and Ethics Committee (See appendix III). Written informed consent was obtained from prospective participants before their participation. The consent form written in simple English clearly stating the purpose, risks, benefits and rights of participants like the right to withdraw any time, in the whole study was used ( See Appendix IV). The participants were assured that their participation was totally voluntary and that if they chose not to participate in the study, it would not affect them, in any way. The nature of commitment in terms of time and form of participation like filling a questionnaire was clearly indicated. Information obtained was treated with utmost confidentiality. No identifiers like name of the participant were indicated on the questionnaire. Only identification numbers and
date for data collection were recorded. Anonymity was further kept by reporting the research findings as group data. Filled questionnaires were kept under lock and key and only accessible to the researcher. Access to data entered on a computer file was through a password known to the researcher only.