The Dementia Quality of Life Scale for Older Family Carers (DQoL-OC) Study

STUDY AIM: This study aims to develop a unique questionnaire to evaluate the quality of life of older people providing care to family members with dementia at home and living in the UK and evaluate its psychometric properties. It is aimed to allow researchers, health and social professionals, as well as support groups' leaders, to quantify the experience of care in old age, for the implementation and evaluation of interventions with these older family carers aiming to improve their QoL. STUDY DESIGN: This study is being carried out through a quantitatively focused sequential and exploratory mixed methods' design (qual ->QUAN) (Creswell and Clark, 2011). It is a multi-centre and convenient sampled study (no randomization). Focus groups were carried out in Nottinghamshire to ask the older family carers what is important for their QoL (REC reference: D10072014 SoHS). The next study phase will be carried out through pilot and re-test studies by using the newly developed DQoL-OC and other validated scales. The study is managed from within the School of Health Sciences at the University of Nottingham.

SELECTION AND WITHDRAWAL OF PARTICIPANTS: The study participants (n=300) will be recruited from GP centres and Memory assessment services. Research posts and leaflets will be available in visible places in the aforementioned NHS settings where family carers usually have access. NHS staff may contact those family carers registered in the NHS system by e-mail in order to inform them about the study. Potential participants will voluntarily contact the PhD researcher Deborah Oliveira by e-mail, work address or telephone number for more information. The research material (questionnaires and PIS) will be posted in a pre-paid envelope to participants' address. About 25 participants taking part in the pilot study will be invited to answer the same questionnaires for a second time, in a re-test session, two weeks after the first set of measures. Participants that accept to take part on the re-test will be sent a second envelope containing the same research material to be filled in and posted back to the researcher within 14 days. This second questionnaire will be sent together with the first one, so no address storage will be necessary. Each participant will be taking part in the study for up to 45 minutes on average. If the participant wants to participate in the re-test phase (2 weeks interval), this amount of time should be doubled. This research will not collect participants' Informed consent (IC), as only questionnaires will be utilized. Completion and subsequent return of questionnaires will be taken as IC and separate written IC will not be sought. Participants taking part in the re-test will receive both sets of measures (two weeks interval) at once and their address information will be immediately destroyed. Participants will be informed that if in 30 days after the pilot questionnaire is received the re-test measures are not received, this will be considered a withdraw from the re-test study and only the pilot measures will be considered for the study.

Pilot study: Research data is collected using a research booklet containing the preliminary version of the DQoL-OC and participants are also invited to comment on the clarity, difficulty and relevance of the overall scale by using Likert scales (1-5), with possible answers ranging from 'very clear' to 'not clear at all' for clarity, 'very difficult' to 'very easy' for levels of difficulty and 'strongly relevant' to 'strongly irrelevant' for relevance, with space for comments in each of these questions. Participants are also invited to comment about the presence of any possible upsetting question. In the same booklet, other self-response questionnaires related to QoL and health are included in order to measure the convergent construct validity of the new scale: the Satisfaction with Life Scale (SWLS) (Diener et al., 1985); the Perceived Health Status Visual Analogue Scale (VAS) (Weinman et al., 1995); an overall HRQoL VAS (Torrance, 1978, Carlsson, 1983, EuroQol, 1990, Bleichrodt and Johannesson, 1997, Lundberg et al., 1999, Shmueli, 1999, Boer et al., 2004, Shmueli, 2005); the World Health Organization Quality of Life Scale for Aging Population - The WHOQOL-AGE (Caballero et al., 2013). The first version of the DQoL-OC had its items created based on the aforementioned qualitative findings and was submitted to an experts' panel to evaluate its relevance and clarity (face and content validity). The preliminary version of the DQoL-OC is divided in three separate sessions totalling 100 questions related to 1) practical aspects of care and caregiving, 2) carers' feelings and concerns and 3) satisfaction with life and caring. All the questions are designed in Likert format (1-5), ranging from "never" to "always", "not at all" to "as much as needed" and "very satisfied" to "very dissatisfied". At the end of the questionnaire there are two open questions that allow carers to write down what most affect their QoL and if there is anything else that was not covered in the questionnaire but yet is important for their QoL.

Re-test: A smaller group of participants (n maximum=25) taking part in the pilot study will be invited by the Researcher to answer the same questionnaires for a second time, in a re-test session, two weeks after the first set of measures. The same procedures utilized in the pilot study will be applied here with regards participant recruitment. Participants that accept to take part on the re-test will be sent a second envelope containing the same research material to be filled in and posted back to the researcher within 14 days. This second envelope will be posted together with the pilot questionnaire and will receive the same code added of a letter for identification and future comparisons. Participants will be informed that they can withdraw of this research activity if they wish to before the 14 days deadline and it is the researcher's responsibility to identify another participant to be included in the re-test study if that is the case. Besides their address details, no other personal data is collected.

DATA ANALYSIS: All the statistic tests will be developed in SPSS 22 statistics package, with confidence intervals of 95% and significant p-value ≤ 0.05. Descriptive (frequencies, mean, median, mode, and standard deviation), suitable correlation statistics (ANOVA) and ceiling and floor effect will be calculated for all the variables. The participants' subjective views regarding the clarity, relevance, levels of difficulty of the new scale DQoL-OC will be analysed by using Kappa coefficient, in order to provide an average score of agreement among the participants about the overall instrument (Streiner and Norman, 1995, Rubio et al., 2003, Leung et al., 2005, Polit and Beck, 2006, Nagpal et al., 2010, Wassef et al., 2014, Landis and Koch, 1977). Aiming to determine whether the DQoL-OC assess the same construct of QoL as other valid and reliable instruments selected from the literature, the construct validity of the DQoL-OC will be evaluated through suitable correlation statistics (ANOVA), correlating the DQoL-OC scores and the other validated scales completed by the participants, as previously mentioned. In addition to that, Factor analysis (FA) will be used to explore the inter relationship between the variables and to further refine and reduce the instrument. For FA, besides the sample size, the strengths of the inter-correlations between the items also must be considered. First of all it is necessary to inspect the correlation matrix to evaluate the suitability of data to FA (should be ≥ 0.30) (Tabachnick and Fidell, 1996). In addition, the extent of Kaiser-Meyer-Oklin of sample adequacy will be performed (which should be ≥ 0.60) (Kaiser and Caffey, 1965, Kaiser, 1970, Dziuban and Shirkey, 1974) and the significance of the Bartlett's Test of Sphericity correlation (Bartlett, 1954) will be analysed to support the factorability of the correlation matrix. According to these results, the data will be submitted to the FA or will be further modified to suitable it to FA. To give an estimate of the equivalence of sets of items from the same instrument, based on the assumption that items measuring the same construct should correlate, the internal consistency of the DQoL-OC will be calculated by using Cronbach's coefficient. Finally, aiming to evaluate whether the measurements provided by the DQoL-OC are stable over the time, the two set of measures of the participants who take part in the re-test will be compared by using Kappa coefficient, with scores ≥ 0.60 for good re-test reliability. The closer the results, the greater the test-retest reliability of the DQoL-OC. The data analysis will take place in UoN computers and will be backed up to the UoN servers.

SAMPLE SIZE AND JUSTIFICATION: There are no agreement in the literature regarding the appropriate sample size to use when conducting a factor analysis (Mundfrom et al., 2005). Even though bigger sample sizes (n>500) are indicative of better factor discrimination (Costello and Osborne, 2005) there are several studies showing that adequate sample size is partly determined by the nature of the data and not only the number of participants, meaning that the stronger the data, the smaller the sample is necessary for an accurate analysis (Fabrigar et al., 1999, MacCallum et al., 1999). Strong data for factor analysis is associated with uniformly high communalities without cross loadings and several variables loading strongly on each other, with such characteristics being considered rare (Costello and Osborne, 2005). Although absolute minimum sample size is not often presented in the literature, in general, small sample sizes are accepted for higher levels of communality within the data and higher ratios of the number of variables to the number of factors (Mundfrom et al., 2005). Mundfrom et al. (2005) stress that small number of variables requires a larger minimum sample size than does a large number of variables, as the relation between the minimum necessary sample size and the number of variables seems compensator, as it does the relation between the minimum necessary sample size and the ratio of the number of variables to the number of factors. These authors propose that for a low communality factors within the matrix under study, a minimum acceptable sample size would be 200 participants. Accordingly, it was stipulated a maximum of 300 participants in this pilot study, considering that the factor communality under study is still unknown and also being careful with the possibility of missing data. The n=25 for the reliability re-test also followed the recommendations available in the literature (Sim and Wright, 2005). Besides the sample size, the strengths of the inter-correlations between the items will be evaluated with the Bartlett test of Sphericity and the extent of Kaiser-Meyer-Oklin (KMO) of sampling adequacy will also be used to assess the adequacy of the data for FA. The data will be then submitted to the FA or will be further modified for suitability for FA. If the study needs a bigger sample size, the Ethics Committee will be contacted and an approval to exceed the number of participants will be sought before the recruitment of more carers.

DATA DISSEMINATION: The findings of the study will be written as part of a PhD thesis in the University of Nottingham with submission date expected for September 2016. The results may be submitted for publication in peer reviewed journals or to academic or scientific events. The data provided by the participants are anonymous and the answers of each participant will receive as specific code that only the Chief investigator and co-investigators will have access. The data may also be used to inform future research and policy. The participants and other stakeholders may receive a fed back of the results in anonymity.