Design
The study employed a multi-phase approach to achieve its purpose. An inductive iterative qualitative process was used to select and develop appropriate scales for pain assessment. The qualitative phase adopted an exploratory descriptive method which allowed identification of features of pain scales and development of a scale which were consistent with the culture of the participants in the study. The philosophy of the qualitative approach was that the perspectives, preferences and interpretations of individuals within a specific socio-cultural context differ [12] and should be incorporated in the development of pain scales. The quantitative approach was used to assess the psychometric properties of the pain scales.
Setting
The first phase of the study was conducted at a tertiary health facility (Korle Bu Teaching Hospital) and a regional hospital (Ridge Hospital) in Accra, Ghana. The second phase was done at the tertiary health facility. The two hospitals have facilities for surgery where various types of surgical procedures are undertaken.
Ethics
The study was approved by the Institutional Review Board of the Noguchi Memorial Institute for Medical Research. Permission was obtained from the authorities of the two hospitals, and individual informed consent obtained from participants both orally and written depending on the participants’ preference and educational background. Anonymity and confidentiality were ensured and participants were assigned identification codes. Identification codes were used to represent the participants: P1-P17 for patients and N1 to N25 for nurses in Phase One and A1-A150 to represent patients in Phase Two.
Sample, procedures and data analysis for the qualitative phase
Nurses and post-operative patients on the surgical wards participated in the study. The inclusion criteria were nurses with a minimum of 6 months working experience on the surgical ward and ambulant patients on the 5th–7th day after surgery who were not in pain. The 6 months clinical experience was considered adequate exposure to enable nurses’ share their experiences on patients’ pain assessment. Patients in pain may have difficulty sharing their experiences because of the pain. Patients and nurses who did not meet the inclusion criteria or give informed consent were excluded.
Purposive sampling technique was employed to recruit participants who met the criteria. The qualitative phase involved individual interviews and focus group discussions. Individual interviews and focus group discussions were used to ensure that ideas for context appropriate pain scale features were generated and discussed from a variety of sources. In contrast to individual interviews, focus groups permitted participants to discuss pain scales. Individual in-depth interviews and focus groups were conducted iteratively to develop and select pain scales. Interviews and discussions were done at a place and time convenient for participants. The participants were allowed to express themselves in the language they were fluent in. Thus, the patients’ interviews were conducted in the local Twi dialect and the nurses’ in English. The interviews and discussions lasted 30 to 45 min. The interviews were recorded and transcribed, while notes were taken during focus group discussions. The first author who is fluent in English and Twi conducted the interviews and moderated the focus groups, and a research assistant took detailed notes during the discussions. This ensured consistency in the data collection process.
The individual interviews involved 18 participants (7 patients and 11 nurses) to generate features of appropriate pain scales. Interviews stopped when data was saturated. During the individual interviews, pain assessment of participants, use of pain scales, and features of pain scales in relation to the context of the study were explored. These were probed in relation to the culture of the patients and nurses. The ideas identified through concurrent content analysis specific to pain scales were discussed during the focus groups. The features generated that were consistent with existing scales, such as numbers and faces pain scales, were selected for discussion during the focus groups. Thus different formats of existing scales were presented and discussed.
In addition, the new culturally specific pain scale features that emerged were developed into scales with appropriate English descriptors for focus group discussions to ensure that the scale and descriptors were consistent with the cultural expression of pain. Participants were actively involved with the choice of English descriptors of the new scale and also discussed those of the existing scales. A research assistant developed all the scales in this study. None of the scales were translated because the primary focus and design of this study was not to translate existing pain scales. We aimed to select, develop and assess the psychometric properties of context appropriate pain scales. The authors are conscious of this limitation and suggest future studies could translate the scales validated in this study using the common local languages in Ghana such as Twi, Ga and Ewe.
Four focus groups were conducted iteratively to develop the new pain scales. The focus groups were made up of two nurses and two patients groups. The number of participants in each nurses’ group was 5 to 7 and consisted of both sexes. The patients’ groups were made up of 5 participants each. Detailed notes were taken during the discussions on recommendations for refinement of the scales. The decision to discard a scale was thus reached by consensus and not by the research team. Different participants were used for the interviews and the discussions to avoid participant fatigue and bias in the development of the scale.
Data management and analysis
Qualitative data was transcribed and read several times to identify the participants’ views on features of pain scales through coding. The notes taken during the focus group discussions were also coded to ensure that any new features of pain scales were incorporated in the analysis. Concurrent content analysis and the iterative process was used to select and develop the appropriate scales. The features identified that were similar to existing scales guided the selection of two existing scales for validation in Phase Two. The data was managed with the NVivo software version 9.
Sample, procedures and data analysis for the quantitative phase: The validation and data collection lasted a period of 4 weeks during which post-operative patients on the surgical and maternity wards of the Korle-bu Teaching Hospital participated. The hospital’s average monthly surgeries of 454 (general surgeries and caesarian sections) was used as the accessible population. With a criterion level of 0.05, we used the Yamane (1967) simplified sample size formula to determine the appropriate sample size of 212. However, 150 patients (70.75 %) met the inclusion criteria during the period of data collection. In the domain of validation or assessing the psychometric properties of pain scales, previous studies have demonstrated that a sample of 150 is acceptable and the type of analysis employed is also acceptable [6, 13, 14]. A census approach was used to recruit participants who met the inclusion criteria for pain assessment. During the clinical validation phase, pain was assessed among post-operative general surgical patients and those who had caesarian sections using three pain scales (0–10 NRS; FPS; and CCPS). Two research assistants (RA1 and RA2) assessed pain with three scales in random order as follows:
-
1.
Inter-rater reliability was assessed at 5 to 10 min intervals between RA1 and RA2 to demonstrate that if different practitioners administered the scales to the same patient or patients with similar pain intensity, consistent results will be obtained
-
2.
Sensitivity to change was assessed by administering the pain scale pre-analgesia and 30 min post-analgesia. Analgesia administration did not take the form of experiment (treatment group and control group) because patients were given different analgesics as prescribed on the ward.
-
3.
Convergent validity was assessed on post-operative days 1, 2 and 3. There was a single assessment at the same time each day because it was assumed that post-operative pain decreased daily after surgery.
Scale preference was assessed by ordering scales–patients indicated their preferred pain scale for assessment in order of preference, after the final pain assessment on post-operative day 3. The assessment criteria were consistent with similar psychometric assessment studies such as the time interval of assessment [15, 16]. It was assumed that changes in pain levels within 5 to 10 min of assessment would not be significant to confound the findings. Participants were shown the scales and were asked to indicate their pain levels. This was done in English or Twi. Twi is the most common local language within the context of study. The two research assistants were both fluent in Twi and were trained by an expert in the language to maintain consistency in administering the scales to patients. The Twi words for pain, translated in Ghana by previous authors [11], were applied in the process. Patients indicated their pain levels orally and the assessors recorded these in English. No observation of pain behaviour was collected in this study.
Statistical analysis was performed using statistical package for the social sciences (SPSS) version 22 (SPSS. Chicago, Illinois, United States of America [USA]). Descriptive statistics were presented in mean and standard deviation (SD), frequency and percentage. Means were compared between groups using Student t test and one-way analysis of variance (ANOVA) for normally distributed variables and Kruskal Wallis was used for non-normally distributed variables. Proportions were compared using Chi-square and Fisher exact test, in case of small cells. Bonferroni test was used to adjust for alpha for multiple pair wise comparison. P-value <0 0.05 was considered statistically significant.
We assessed the following psychometric properties of the scales
Convergent validity by Pearson’s correlation coefficient, inter-rater reliability by interclass correlation coefficient (ICC) [17], sensitivity to change in pain intensity following analgesia and with increasing duration after surgery (post-operative day 1 to 3) using repeated measures of ANOVA [18].
Copyright permission was sought from Elsevier to use the Wong-baker FACES pain scale (Wong & Baker, 2001) for clinical pain assessment. The 0–10 NRS is a free access pain scale. Identification codes used for scales were: Scale 1 = 0–10 NRS (S1); Scale 2 = Wong-Baker Faces Pain Scale (S2); Scale 3 = Colour-Circle pain Scale (S3).
Trustworthiness for qualitative phase
Processes undertaken to ensure rigor or validity and reliability included effective probing to ensure that participants’ ideas on features of pain were fully explored. Moderation of focus group discussions and interviews were conducted by the first author who is experienced in qualitative data collection. Concurrent and iterative data collection and analysis were done to ensure identification and development of pain scales that were appropriate for the context of the study. All the authors were involved in data verification and analysis to ensure that participants’ views were captured. Detailed audit trail was maintained so that future researchers can confirm the processes used in this study.
Validity and reliability for quantitative phase
Patients who were involved in the scale development and selection did not participate in the clinical validation of the scales as they had been discharged at the time of validation. Also, two research assistants assessed pain for all the patients in this study to avoid participant familiarity, which could confound the data generated especially where the assessor demonstrated negative attitude during pain assessment. Quantitative data was checked for completeness and omissions before analysis to avoid error in analysis.