Efficacy and experiences of telephone counselling for informal carers of people with dementia (Review)

Transcription

1 Efficacy and experiences of telephone counselling for informal carers of people with dementia (Review) Lins S, Hayder-Beichel D, Rücker G, Motschall E, Antes G, Meyer G, Langer G This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library 2014, Issue 9

2 T A B L E O F C O N T E N T S HEADER ABSTRACT PLAIN LANGUAGE SUMMARY SUMMARY OF FINDINGS FOR THE MAIN COMPARISON BACKGROUND OBJECTIVES METHODS Figure RESULTS Figure Figure Figure Figure Figure Figure DISCUSSION AUTHORS CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES CHARACTERISTICS OF STUDIES DATA AND ANALYSES Analysis 1.1. Comparison 1 Telephone counselling, Outcome 1 Depressive symptoms Analysis 1.2. Comparison 1 Telephone counselling, Outcome 2 Burden Analysis 1.3. Comparison 1 Telephone counselling, Outcome 3 Support ADDITIONAL TABLES APPENDICES CONTRIBUTIONS OF AUTHORS DECLARATIONS OF INTEREST SOURCES OF SUPPORT DIFFERENCES BETWEEN PROTOCOL AND REVIEW i

3 [Intervention Review] Efficacy and experiences of telephone counselling for informal carers of people with dementia Sabine Lins 1, Daniela Hayder-Beichel 1, Gerta Rücker 2, Edith Motschall 2, Gerd Antes 1, Gabriele Meyer 3, Gero Langer 4 1 German Cochrane Centre, Institute of Medical Biometry and Medical Informatics, University Medical Center Freiburg, Freiburg, Germany. 2 Center for Medical Biometry and Medical Informatics, Medical Center - University of Freiburg, Freiburg, Germany. 3 Institute of Health and Nursing Sciences, Martin-Luther-University Halle-Wittenberg, Halle (Saale), Germany. 4 Institute of Health and Nursing Sciences, German Center for Evidence-based Nursing, Martin Luther University Halle-Wittenberg, Halle (Saale), Germany Contact address: Sabine Lins, German Cochrane Centre, Institute of Medical Biometry and Medical Informatics, University Medical Center Freiburg, Berliner Allee 29, Freiburg, 79110, Germany. lins@cochrane.de. Editorial group: Cochrane Dementia and Cognitive Improvement Group. Publication status and date: New, published in Issue 9, Review content assessed as up-to-date: 25 February Citation: Lins S, Hayder-Beichel D, Rücker G, Motschall E, Antes G, Meyer G, Langer G. Efficacy and experiences of telephone counselling for informal carers of people with dementia. Cochrane Database of Systematic Reviews 2014, Issue 9. Art. No.: CD DOI: / CD pub2. Background A B S T R A C T Informal carers of people with dementia can suffer from depressive symptoms, emotional distress and other physiological, social and financial consequences. Objectives This review focuses on three main objectives: To: 1) produce a quantitative review of the efficacy of telephone counselling for informal carers of people with dementia; 2) synthesize qualitative studies to explore carers experiences of receiving telephone counselling and counsellors experiences of conducting telephone counselling; and 3) integrate 1) and 2) to identify aspects of the intervention that are valued and work well, and those interventional components that should be improved or redesigned. Search methods The Cochrane Dementia and Cognitive Improvement Group s Specialized Register, The Cochrane Library, MEDLINE, MEDLINE in Process, EMBASE, CINAHL, PSYNDEX, PsycINFO, Web of Science, DIMDI databases, Springer database, Science direct and trial registers were searched on 3 May 2011 and updated on 25 February A Forward Citation search was conducted for included studies in Web of Science and Google Scholar. We used the Related Articles service of PubMed for included studies, contacted experts and hand-searched abstracts of five congresses. 1

4 Selection criteria Randomised controlled trials (RCTs) or cross-over trials that compared telephone counselling for informal carers of people with dementia against no treatment, usual care or friendly calls for chatting were included evaluation of efficacy. Qualitative studies with qualitative methods of data collection and analysis were also included to address experiences with telephone counselling. Data collection and analysis Two authors independently screened articles for inclusion criteria, extracted data and assessed the quantitative trials with the Cochrane Risk of bias tool and the qualitative studies with the Critical Appraisal Skills Program (CASP) tool. The authors conducted metaanalyses, but reported some results in narrative form due to clinical heterogeneity. The authors synthesised the qualitative data and integrated quantitative RCT data with the qualitative data. Main results Nine RCTs and two qualitative studies were included. Six studies investigated telephone counselling without additional intervention, one study combined telephone counselling with video sessions, and two studies combined it with video sessions and a workbook. All quantitative studies had a high risk of bias in terms of blinding of participants and outcome assessment. Most studies provided no information about random sequence generation and allocation concealment. The quality of the qualitative studies ( thin descriptions ) was assessed as moderate. Meta-analyses indicated a reduction of depressive symptoms for telephone counselling without additional intervention (three trials, 163 participants: standardised mean different (SMD) 0.32, 95% confidence interval (CI) 0.01 to 0.63, P value 0.04; moderate quality evidence). The estimated effects on other outcomes (burden, distress, anxiety, quality of life, self-efficacy, satisfaction and social support) were uncertain and differences could not be excluded (burden: four trials, 165 participants: SMD 0.45, 95% CI to 0.90, P value 0.05; moderate quality evidence; support: two trials, 67 participants: SMD 0.25, 95% CI to 0.73, P value 0.32; low quality evidence). None of the quantitative studies included reported adverse effects or harm due to telephone counselling. Three analytical themes (barriers and facilitators for successful implementation of telephone counselling, counsellor s emotional attitude and content of telephone counselling) and 16 descriptive themes that present the carers needs for telephone counselling were identified in the thematic synthesis. Integration of quantitative and qualitative data shows potential for improvement. For example, no RCT reported that the counsellor provided 24-hour availability or that there was debriefing of the counsellor. Also, the qualitative studies covered a limited range of ways of performing telephone counselling. Authors conclusions There is evidence that telephone counselling can reduce depressive symptoms for carers of people with dementia and that telephone counselling meets important needs of the carer. This result needs to be confirmed in future studies that evaluate efficacy through robust RCTs and the experience aspect through qualitative studies with rich data. P L A I N L A N G U A G E S U M M A R Y Efficacy and experiences of telephone counselling for informal carers of people with dementia What is the problem? Caring for a person with dementia often has a negative impact on an informal carer s mental and physical health and social life. Therefore, these informal carers should be offered support. They are usually family members and care for the person with dementia at home. What is telephone counselling? Eliciting a person s concerns, listening, and providing support, information, or teaching in response to a persons s stated concerns, over the telephone. The purpose of this review The aim of this review was to investigate whether telephone counselling is an effective way of reducing symptoms of depression and other stresses in the carers of people with dementia. We also investigated which aspects of telephone counselling the people who received it thought could be improved. Results of the review 2

5 We searched for scientific studies that compared telephone counselling with no treatment, or usual care or friendly calls. We found nine studies that investigated how well telephone counselling worked (efficacy) and two studies that examined the quality of the experience. The efficacy studies investigated three types of telephone counselling: telephone counselling only (six studies); telephone counselling plus video sessions (one study); and telephone counselling plus video sessions and a workbook (two studies). These provided some evidence that telephone counselling is effective for reducing depressive symptoms in carers of people with dementia (three studies), but no clear positive effects could be shown for any other outcome such as stress or anxiety. The studies that investigated the experience aspect of telephone counselling revealed a range of carers needs (16 themes) that covered three main areas: barriers to - and things that enabled - successful implementation of telephone counselling; the counsellor s emotional attitude; and the content of the telephone counselling. All studies were of moderate quality. Conclusions Analysis of both sets of results, i.e. efficacy compared with information about carers experiences of telephone counselling, revealed needs that so far have not been met by telephone counselling. The studies that examined the experience aspect covered a very limited range of telephone counselling. The results of this review should be interpreted with caution due to the small number of included studies and their moderate quality. 3

6 S U M M A R Y O F F I N D I N G S F O R T H E M A I N C O M P A R I S O N [Explanation] Telephone counselling for informal carers of people with dementia Patient or population: informal carers of people with dementia Settings: Intervention: telephone counselling Outcomes Illustrative comparative risks*(95% CI) Relative effect (95% CI) Depressive symptoms Burden Assumed risk Control Corresponding risk Telephone counselling The mean depressive symptoms in the intervention groups was 0.32 standard deviations lower (0.63 to 0.01 lower) The mean burden in the intervention groups was 0.45 standard deviations lower (0.90 lower to 0.01 higher) No of participants (studies) 163 (3 studies) 165 (4 studies) Quality of the evidence (GRADE) moderate 1,2 moderate 3,4,5 Comments SMD0.32(0.01to0.63) SMD0.45(-0.01to0.90) Support The mean support in the intervention groups was 0.25 standard deviations higher (0.24 lower to 0.73 higher) 67 (2 studies) low 6,7,8 SMD0.25(-0.24to0.73) 4

7 *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention(and its 95% CI). CI: Confidence interval; SMD: standardised mean difference GRADE Working Group grades of evidence High quality: further research is very unlikely to change our confidence in the estimate of effect Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate Lowquality:furtherresearchisverylikelytohaveanimportantimpactonourconfidenceintheestimateofeffectandislikelytochangetheestimate Very low quality: we are very uncertain about the estimate 1 Allthreestudies:noITTanalysis 2 Allthreestudies:blindingwasnotpossibleduetotheinterventionandtheself-reportedoutcomemeasures 3 Allfourstudies:noITTanalysis 4 Allfourstudies:blindingwasnotpossibleduetotheinterventionandtheself-reportedoutcomemeasures 5 Onlytwooutoffourstudiesreportedthesignificantoutcomes 6 Bothstudies:noITTanalysis 7 Bothstudies:blindingwasnotpossibleduetotheinterventionandtheself-reportedoutcomemeasures 8 Toofewparticipants 5

8 B A C K G R O U N D Description of the condition Caring for people with dementia is a challenge for the people involved and for society as a whole. In 2013, it was estimated that million people worldwide suffered from dementia; 17 million of these were living in high-income countries and 27 million were living in low and middle-income countries (Prince 2013). It can be assumed that every person receiving medical/social care for dementia who is living at home has at least one informal carer (Brodaty 2003). Informal carers of people with dementia are on average 55 years old and 65% are female. More than 50% are a child of the person with dementia and 28% are the spouse (Ferrario 2003). Caring often leads to negative psychological, physiological and financial consequences for these informal carers (Baumgarten 1992; Max 1995; Pinquart 2003). Many of them show high, or moderate to severe, levels of burden (Brodaty 2013). A long caring period results in physical discomfort; though if people with dementia have functional restrictions and show behavioural syndromes, informal carers suffer more often from depression and physical discomfort (Baumgarten 1992). It has been estimated that 30% of informal carers suffer from depressive symptoms (Taylor 2008), with 62% of women and 52% of men stating they suffer emotional stress (Alzheimer s Association 2014). Half of the informal carers are likely to give up their holidays, hobbies or social contacts; and 18% remarked that caring harmed their health (Alzheimer s Association 2004). Informal male carers deal with care-related burdens differently from women (Almberg 1998; Larranaga 2008). Men perceive a lack of positive prospects and need support. Women, on the other hand, report an increased strain in their relationships with other family members and an increase in their own health problems. However, caregiver burden can lead to early transfer of the person with dementia to a nursing home (Yaffe 2002), which should usually be avoided. Therefore, informal carers of people with dementia should be offered support opportunities. Throughout this review the term carer always refers to informal carers providing home care for people with dementia. Description of the intervention The use of telecommunication for healthcare purposes is widespread. Telephone counselling is one way of using this technology, and is an attractive option for people who do not have the opportunity to participate in traditional face-to-face counselling (Reese 2006). Occupational groups such as nurses or psychotherapists use telephone counselling for a wide variety of diseases and situations. In the 5 th Nursing Interventions Classification telephone consultation is defined as: Eliciting a patient s concerns, listening, and providing support, information, or teaching in response to a patient s stated concerns, over the telephone (College of Nursing 2005). How the intervention might work Reese, and colleagues, investigated the use of telephone counselling for a variety of issues such as depression, anxiety, relationship problems, grief or work-related difficulties (Reese 2002). Participants described telephone counselling as helpful for global and specific improvements, and were satisfied with the counselling they received. The efficacy of telephone counselling was also confirmed by the results of a study on smoking cessation (Stead 2013). Since the carers of people with dementia frequently suffer from depressive symptoms, in theory, they could also benefit from telephone counselling. Why it is important to do this review Informal carers are an important care resource for people with dementia, so interventions that support them and reduce their burden should be available. Telephone support practised in clinical settings is one option that might reduce the burden. Only one narrative review of limited quality has been published on telephone counselling for carers of people with dementia ( Mason 2008), so there is a need for a high quality and up-to-date systematic review. The results of this review could show that telephone counselling might be a valid alternative to face-to-face interventions in rural areas where healthcare professionals have to travel long distances to reach the carers. Alongside a summary of randomised controlled trials (RCTs) addressing the question of efficacy, we use qualitative evidence to address peoples experiences with the intervention. Carers perspectives and the range of carers needs, for example, may help in understanding whether the carers find telephone counselling helpful, but miss the face-to-face contact because of their limited social contacts. Study findings indicate that telephone support is not helpful for some people (Chang 2004). A synthesis of qualitative evidence may help explain why certain people benefit more than others and provide more detailed information about the context, characteristics and circumstances of this group. An earlier Cochrane review on the efficacy of cognitive reframing for carers of people with dementia did not consider the experiences of the carers of people with dementia (Vernooij-Dassen 2011). It included interventions aimed at cognitive reframing but excluded interventions aimed at practical support. By comparison, our review includes telephone counselling that gives practical advice, but excludes interventions delivered by ongoing face-toface counselling. In contrast to the earlier review (Mason 2008), we conducted a comprehensive and systematic search strategy, assessed the methodological quality of the included studies, and syn- 6

9 thesised the evidence for both aspects - efficacy and experience - separately. In order to provide clear differentiation of the two aspects being investigated, each section in this review consists of two subsections: To evaluate efficacy and To evaluate experience, or, where applicable, To evaluate efficacy and experience. O B J E C T I V E S This review focuses on three main objectives: To: 1) produce a quantitative review of the efficacy of telephone counselling for informal carers of people with dementia; 2) synthesize qualitative studies to explore carers experiences of receiving telephone counselling and counsellors experiences of conducting telephone counselling; and 3) integrate 1) and 2) to identify aspects of the intervention that are valued and work well, and those interventional components that should be improved or redesigned. M E T H O D S Criteria for considering studies for this review Types of studies Evaluation of efficacy We included all individually randomised, parallel-group controlled trials that compared telephone counselling for carers of people with dementia with no treatment, usual care or friendly calls for chatting. We also considered cross-over trials. Evaluation of experience We included studies that used qualitative methods of data collection and analysis, that is, either a stand-alone study or a discrete part of a larger mixed-method study. Also, we considered qualitative elements from process evaluations conducted alongside trials. Types of participants Evaluation of efficacy and experience We included informal carers who provide care for a person with dementia at home without being reimbursed for their caring work. The adult carer was a relative, a friend or a neighbour of the person with dementia. There were no restrictions regarding sex, age or ethnic background. In order to evaluate experience, counsellors experiences of telephone counselling were considered and involved in the synthesis as well, if their experiences were investigated in combination with caregivers experiences. Types of interventions Evaluation of efficacy and experience Telephone counselling is a complex intervention. It is also used in combination with face-to-face consultations or computer-involved interventions. For this review, we considered all telephone counselling interventions that did not include any face-to-face counselling (a face-to-face contact at the first meeting was acceptable) that addressed informal carers of people with dementia living in the community. The telephone counselling was not allowed to be part of another complex intervention. It had to be conducted for a minimum of two months, but we made no restrictions about the frequency, length or number of telephone calls. The telephone counselling had to be performed by a health professional, such as a nurse or a psychologist, with expertise in dementia. Telephone counselling had to comprise at least three components: 1. general information about dementia; 2. educating of carers in coping skills and caring for their own health; 3. psychosocial support where carers could share their feelings and were shown how to build up their social network. Types of outcome measures Evaluation of experience We included all experiences regarding the intervention of interest. Evaluation of efficacy We included all studies that had either the following primary or secondary outcome(s). 7

10 Primary outcomes Electronic searches Evaluation of efficacy The primary outcomes were depressive symptoms, since they are frequent among informal carers of people with dementia (Baumgarten 1992). Secondary outcomes Evaluation of efficacy Carer-related outcomes: 1. caregiver burden; 2. distress; 3. anxiety; 4. quality of life; 5. care-giving self-efficacy; 6. satisfaction. Outcomes concerning the person with dementia: 1. institutionalisation; 2. mood; 3. quality of life. Search methods for identification of studies Evaluation of efficacy and experience The efficacy of and experience with telephone counselling address the same intervention and population in the same situation, but to answer these questions, both quantitative and qualitative study designs are needed. We searched databases focusing on the intervention (telephone counselling) and the population (carers of people with dementia) without a limitation regarding study designs. We expected that retrieved references would include both quantitative and qualitative studies. We did not identify too many references, so we did not have to limit the search results by quantitative or qualitative filters. Our search strategy for efficacy (quantitative trials) followed Cochrane Collaboration guidelines (Lefebvre 2011). For the question of experience (qualitative studies), we followed advice in Chapter 20 (Noyes 2011a), section , which provides supplemental guidance on searching for qualitative studies supplied by the Cochrane Qualitative and Implementation Methods Group. Details of the full search, with all search strategies and the number of records identified in each database is included in the Appendices 1 to 22 of the review. Evaluation of efficacy and experience We searched ALOIS ( the Cochrane Dementia and Cognitive Improvement Group s Specialized Register. Our approach was to assess all carer-focused studies within ALOIS. ALOIS is maintained by the Trials Search Co-ordinator of the Cochrane Dementia Group and contains dementia and cognitive improvement studies identified from: 1. monthly searches of a number of major healthcare databases: MEDLINE, EMBASE, CINAHL, PsycINFO and LILACS; 2. monthly searches of a number of trial registers: metaregister of Controlled Trials (mrct); UMIN (Japan Trial Register); ICTRP (the World Health Organization (WHO) portal that covers ClinicalTrials.gov; ISRCTN; Chinese Clinical Trials Register; German Clinical Trials Register; Iranian Regsitry of Clinical Trials and the Netherlands National Trials Regsiter, plus others); 3. quarterly search of The Cochrane Library s Central Register of Controlled Trials (CENTRAL); 4. six-monthly searches of a number of grey literature sources: ISI Web of Knowledge Conference Proceedings; Index to Theses; Australasian Digital Theses. To view a list of all sources searched for ALOIS see About ALOIS on the ALOIS website. We ran additional searches in many of the above plus the following sources to ensure that both qualitative studies and the most upto-date results of quantitative trials were retrieved. An overview of the search results is displayed in Appendix 1. The search strategies used can be seen Appendix 2, Appendix 3, Appendix 4, Appendix 5, Appendix 6, Appendix 7, Appendix 8, Appendix 9, Appendix 10, Appendix 11, Appendix 12, Appendix 13, Appendix 14, Appendix 15, Appendix 16, Appendix 17, Appendix 18, Appendix 19, Appendix 20, Appendix 21, and Appendix 22. We also searched the following: MEDLINE (Ovid SP); MED- LINE in Process & other non-indexed citations (Ovid SP); The Cochrane Library (Wiley Online Library); CINAHL (EBSCOhost); PSYNDEX (EBSCOhost); PsycINFO (EBSCOhost); EM- BASE (DIMDI); Web of Science (Thomson Reuters) (Science Citation Index Expanded (SCI-EXPANDED), Social Science Citation Index (SSCI), Arts & Humanities Citation Index (A&HCI), Conference Proceedings Citation Index-Science (CPCI-S), Conference Proceedings Citation Index Social Science & Humanities (CPCI-SSH)); DIMDI databases: Global Health, CCMed, DAHTA-Database, HECLINET (1969 to 2001), ISTPB + ISTP/ ISSHP, SOMED (1978 to 2000), Health Technology Assessment, and Publishers databases (up to 2011): Hogrefe, Karger, Kluwer, Krause & Pachernegg, Thieme; Springer (via Springerlink); Science Direct (Elsevier). HECLINET and SOMED were 8

11 only searched up to 2001/2000 as the databases ceased to be updated then. The Publishers databases Hogrefe, Karger, Kluwer, Krause & Pachernegg and Thieme were searched up to 2011 since in 2013 we had no access to these databases. Data extraction and management We performed data extraction differently for the evaluations of efficacy and experience and considered quantitative and qualitative study designs. Searching other resources Evaluation of efficacy and experience We handsearched abstracts of the following congresses (see Table 1): International Conference of Alzheimer s Disease International; Alzheimer Europe Conference; Dementia Services Development Centre International Conference; Dementia Congress; and International Congress on Vascular Dementia. We also searched the trial registries of ClinicalTrials.gov; Controlled Trials and the WHO. We checked the references of included studies and conducted a Forward Citation Search of these studies in Web of Science and Google Scholar. We also used the Related Articles service of PubMed for the included studies to find further relevant trials. The investigators of the included RCTs were contacted to ask for relevant quantitative studies and qualitative studies that were associated with, or conducted alongside, the effectiveness studies. Data collection and analysis Selection of studies Evaluation of efficacy and experience We transferred the references identified into reference management software (Endnote). Two independent authors (SL, GM) reviewed the titles and abstracts. Based on the full texts, two unblinded authors independently examined whether the studies fulfilled the inclusion criteria. In the preliminary search, the authors became acquainted with the names and institutions of study authors and the journal of publication of most of the relevant articles, so a blinded assessment of articles relevance was no longer possible. We resolved disagreement by discussion with a third review author (GL). We linked multiple articles from the same study together. We have listed studies that initially seemed to meet the eligibility criteria but on further inspection were seen no to, and studies that did not meet all of the criteria, with the reasons for their exclusion (see Characteristics of excluded studies). Evaluation of efficacy We based the data extraction form for the quantitative studies on the recommended items from the Cochrane Handbook, Chapter 7 (Higgins 2011), section The form covered information on the authors; publication date; country; study design; characteristics of the study population including setting; inclusion and exclusion criteria; credentials of those who provided the treatment; treatment type, duration, intensity, frequency; control intervention and outcomes. We conducted a pre-test using two randomlyselected studies and adapted the form afterwards. Two researchers (SL and DHB or AG or CH or JR) extracted data independently. We resolved disagreement by discussion. We requested missing data or further information related to the risk of bias assessment from the original investigators. We contacted eight authors of included studies, but only four responded; three authors provided further information for our risk of bias assessment, and two provided numerical data. We documented all essential steps in data extraction as requested in the PRISMA statement (Liberati 2009). Evaluation of experience The data extraction form included the following items: trial authors, publication date, country, methodology, methods, phenomena of interest, characteristics of the study population including setting, inclusion and exclusion criteria, credentials of those who provided the study, intervention outcomes, methods of data collection and analysis, and findings such as data from the participants and comments of study authors. One researcher (SL) created the data extraction form and a manual for its use prior to the data extraction. The data extraction form was adapted slightly during data extraction of the first and second studies and checked by another researcher (DHB). One researcher (SL) extracted the data on participants quotes and study authors commentaries. A second researcher (DHB) checked the extracted data. Disagreement was resolved by discussion. The data extraction and other aspects of synthesis were reported according to the ENTREQ statement (Tong 2012). Assessment of risk of bias in included studies Evaluation of efficacy We performed critical appraisal of the risk of bias of included studies using The Cochrane Collaboration s tool for assessing risk of bias. Two researcher (SL and GL) performed the assessment independently. Blinding to the names of the authors, institutions, 9

12 journal and results of a study was not feasible because most of the studies were known to the review authors after the first specific literature search. We resolved disagreement by discussion and consulted a third researcher (GM) where necessary. We requested further information from the original investigators. Evaluation of experience When assessing the methodological quality of qualitative studies, it is important to evaluate researcher bias as a core criterion (cf Hannes 2011). In qualitative studies researchers have an impact on the data, so they must show their ability and the efforts they undertook to prevent this from influencing the data. Researchers have to state that the process was rigorous and trustworthy so that credibility of the findings can be assumed. The appraisal of qualitative research includes three stages (Hannes 2011): 1. filtering the identified hits according to the inclusion and exclusion criteria; 2. technical appraisal with evaluation of the credibility, transferability, dependability and confirmability of the original research papers included; 3. judging the methodological coherence. For this, the congruity between paradigms that guide the research project and the methodology and methods chosen is evaluated. The Cochrane Qualitative and Implementation Methods Groups recommends the third stage whenever the instrument for critical appraisal does not cover a paradigmatic approach. We used the Critical Appraisal Skills Program (CASP) for judgment of the qualitative studies included. CASP does not include a question relating to the paradigm that guides the study. Therefore an evaluation of methodological coherence was conducted. Two researchers assessed study quality. One (SL) performed the critical appraisal and another researcher (DHB), with experience in qualitative research, checked the judgments. The consequence of the critical appraisal of the qualitative research can be a weighting of the studies found in the search. High quality studies can be given more weight than those of low quality. At present there are no established parameters to determine the weight of qualitative studies. To enhance the applicability of the findings of qualitative synthesis, we used the CerQual (confidence in the qualitative evidence) approach (Glenton 2013). The CerQual approach is similar to GRADE (Grading of Recommendations Assessment, Development and Evaluation) and considers the methodological quality of evidence and the coherence of synthesised findings. For appraisal of methodological limitations, we used the results of the CASP assessment. For assessment of coherence, we determined the extent to which synthesised qualitative findings were grounded in the data of individual studies. If a synthesised finding was grounded across individual studies from different settings, then coherence was considered to be increased. Measures of treatment effect To evaluate efficacy In meta-analysis, longer ordinal scales are often analysed as continuous data. Depression scales can have different types of outcome data. For continuous data measured by different psychometric scales, we calculated the standardised mean difference. If only one assessment had been used in the trials, the weighted mean difference would have been applied. The scales were checked with regard to whether or not they had been validated and whether the original, or an adapted questionnaire had been used. Due to the expected heterogeneity, pooling was done using a random-effects model. Pooling of some studies was not possible, so the results of those are presented narratively. We used the standard software of The Cochrane Collaboration (Review Manager 5.2, RevMan 2012) for analysis and graphical display. We used the open statistical environment R (version ) for extracting standard deviations from P values (R 2012). Dealing with missing data Evaluation of efficacy We evaluated the missing data and drop-out rates for each RCT and displayed the results in Characteristics of included studies. Missing data could concern outcomes, summary data, individuals or, in the case of subgroup analysis or meta-regression, study-level characteristics. We requested all missing data from the original investigators and a sensitivity analysis was performed. Assessment of heterogeneity Evaluation of efficacy We assessed heterogeneity by Cochran s Q test and quantified it by using the I² statistic. We used a random-effects model for pooling data. Where there was pronounced heterogeneity (i.e. I² > 60%) we planned to explain heterogeneity by subgroup analyses or metaregression as described under Subgroup analysis and investigation of heterogeneity. Assessment of reporting biases Evaluation of efficacy We took steps to detect reporting bias. We checked each study to see whether outcomes were mentioned in the protocol, the methods section and the results sections of the report and the conference presentation, if available. We asked the authors whether a published protocol was available. If outcomes mentioned in the 10

13 protocol or in the methods section of the report were not presented in the results section of the report or the conference presentations, we contacted the study authors. Data synthesis First synthesis: evaluation of efficacy In the absence of clinical and statistical heterogeneity, we conducted a meta-analysis using the random-effects model in Review Manager 5.2. The separate control interventions were analysed in separate meta-analyses. Second synthesis: evaluation of experience We conducted thematic synthesis for qualitative data. Such a synthesis can produce new insights and understanding as studies with different important aspects, such as population and settings, can be combined to identify concepts that are present across a range of contexts and settings. The choice of synthesis methods depends on several aspects (see the supplemental Handbook guidance of the Cochrane Qualitative and Implementation Methods Group; Noyes 2011b), and the review question and purpose should be considered. In our case, this concerns the carers and counsellors experiences of telephone counselling. Another aspect to be considered is the identified evidence: if descriptive qualitative studies that are characterised by thin descriptions are available, an aggregative method is more appropriate; if the identified evidence includes a high proportion of in-depth qualitative studies characterised by thick descriptions, an interpretative method can be applied. Authors should also consider whether or not an appropriate theoretical model for the phenomenon of interest exists. Since we only identified two descriptive studies that were characterised by thin descriptions with no appropriate theoretical model, we used thematic synthesis as described by Thomas We transferred the extracted findings of the studies into the qualitative data analysis software MAXQDA Version The first step of the analysis was an intensive reading of the included studies during which the initial ideas and memos were recorded in a researcher s diary. This step was followed by a lineby-line coding, conducted in the review authors mother language of German, to describe the content and meaning of the data with the review author s own words. During the line-by-line coding, ideas for grouping the data and descriptive themes were noted in the researcher s diary. The third step included summarising codes into descriptive themes by looking for similarities and differences between the codes. This step was influenced by the first idea on possible analytical themes and a structure which displays how they can be related to each other. Afterwards the descriptive themes were grouped into analytical themes, and the first idea of the relationship between the analytical themes was modified. In this step the analytical themes were checked and refined by looking at the underlying descriptive theme, codes and original data of the studies. We discussed the preliminary findings with another author with experience in qualitative research. We considered the descriptive and analytical themes and their distinctions from each other, afterwards some thematic themes were grouped together and reworded. We discussed and modified the specifications of the descriptive and analytical themes. We also adapted the structure or the relationship between the analytical themes in terms of a mapping. At the end of this process, we derived three analytical themes that were based on 16 descriptive themes (see Figure 1) and 119 codes. There was, however, little opportunity for synthesis with only five themes shared between the two studies. Additional limiting factors concerning the included qualitative evidence are described under the Included studies section below. 11

14 Figure 1. Aspects of telephone counselling for informal carers of people with dementia Third synthesis: integration of the first synthesis and the product of the second synthesis The aim of the third synthesis was to merge the question of efficacy (quantitative studies) with the question of experience (qualitative studies) while juxtaposing them in a matrix (Thomas 2004). The integration of qualitative and quantitative data was guided by two questions. 1. Which themes of experience are addressed by the interventions of the RCTs? 2. Which aspects of the RCT interventions were not mentioned by the participants? The first question was captured in the first part of the integration of qualitative and quantitative data within Matrix 1 (Table 2), where the descriptive themes from the qualitative data were juxtaposed with the RCTs. This first comparison should reveal which descriptive themes were already considered in the telephone counselling intervention of the RCTs and which were missing, and also identify those interventions that contain more items of positive experience than others. Consequently, these findings can be used for subgroup analyses. The second question was captured in the second part of the integration of qualitative and quantitative data within Matrix 2 (Table 3). Here the results of the telephone counselling intervention of the RCTs were juxtaposed with the qualitative studies. The second comparison made clear which aspects of telephone counselling were already considered by the qualitative studies and which were not considered. This matrix is interesting for appraising the range of findings of the qualitative studies; in addition, aspects that may be less important for the carers were indicated. Subgroup analysis and investigation of heterogeneity Evaluation of efficacy If five studies for each planned aspect had been available, subgroup analysis or meta-regression would have been conducted for the following areas. 1. Carer-related: relation to the person with dementia (e.g. spouse, child); support by other family members; pre-existing stress of the carer; in employment or not; gender. 2. Intervention-related: duration; intensity of intervention. 3. Disease-related: stage of dementia. 4. Experience-related: aspects of the intervention that were experienced positively. 12

15 Sensitivity analysis Evaluation of efficacy We applied sensitivity analyses to check the impact of different ways of handling missing data as described in Dealing with missing data. R E S U L T S Description of studies Nine quantitative studies and two qualitative studies met the inclusion criteria. The characteristics of all included studies are described in Characteristics of included studies. Results of the search Evaluation of efficacy and experience The first search was performed in May 2011 and an update search in February The check for duplicates of records identified by database searching was first conducted electronically using reference management software and then conducted by hand. The records identified by searching non-electronic sources could not be transferred into reference management software. An electronic check for duplicates of those records was therefore not possible. The titles and abstracts of the 7877 retrieved records were independently screened by two authors (SL and DHB or academic support staff). The remaining 110 records were independently assessed for inclusion on the basis of full texts by two review authors (SL and GM or DHB). Thirteen publications fulfilled the inclusion criteria (for efficacy: Chang 1999; Davis 2004; Finkel 2007, Gant 2007; Glueckauf 2007; Steffen 2000; Tremont 2008; Wilz unpublished; Winter 2007; for experience: Chang 2004; Salfi 2004; Salfi 2005). Salfi 2004 (Thesis) and Salfi 2005 (article) refer to the same study. Wilz 2011 published the results of one outcome and recorded other outcome results in a manuscript that has been submitted for publication (Wilz unpublished). Two ongoing trials were identified: one through searching trial registries (Nichols 2005), and one as a poster abstract (Tremont 2011). In total, we included nine studies that addressed the efficacy and two studies that addressed the experience of telephone counselling (see Figure 2). 13

16 Figure 2. Flow diagram of study selection process 14

17 Included studies Evaluation of efficacy All the included quantitative trials were RCTs; all except one were conducted in the USA, the exception was performed in Germany (Wilz unpublished). Seven studies included between 32 and 71 carers of people with dementia (Chang 1999; Davis 2004; Finkel 2007; Gant 2007; Glueckauf 2007; Steffen 2000; Tremont 2008), one study included 103 carers (Winter 2007), and another included 229 (Wilz unpublished). Telephone counselling The interventions in the included studies addressing efficacy differed. In six studies, telephone counselling was provided without an additional intervention (Davis 2004; Finkel 2007; Glueckauf 2007; Tremont 2008; Wilz unpublished; Winter 2007). One study combined telephone counselling with an educational video-series (Chang 1999), and two studies combined it with educational video-series plus a workbook (Gant 2007; Steffen 2000). Since we had decided to include only trials that conducted counselling, we assessed the included studies for whether they provided counselling or therapy. It is difficult to differentiate between counselling and therapy since the boundary between the concepts is blurred. Usually the target group for counselling does not need therapy, and telephone counselling will be performed for a shorter period than therapy. We therefore tried to demarcate the concepts on the basis of the intervention s duration, theoretical background and the participants condition. The theoretical background for telephone counselling gives an impression of the way in which the counselling was performed and whether it should be placed into the area of counselling or fits better into the area of therapy. If the theoretical background was not mentioned in the publication, study investigators were contacted and asked to explain it. Table 4 lists the included studies with their intervention, duration, theoretical background and the review authors decisions on whether the intervention was to be allocated to the area of counselling or therapy. Table 5 gives an overview of the treatment and control group details of the included quantitative trials. Most of the included studies conducted telephone counselling on an individual basis (Chang 1999; Davis 2004; Gant 2007; Steffen 2000; Tremont 2008; Wilz unpublished). Two studies combined group sessions over the telephone with individual sessions (Finkel 2007; Glueckauf 2007). One study conducted all the telephone counselling via group sessions (Winter 2007). Duration of telephone counselling varied amongst the included studies. Two studies were conducted over two months (Chang 1999; Steffen 2000), four studies over three months (Davis 2004; Gant 2007; Glueckauf 2007; Wilz unpublished), two studies over six months (Finkel 2007; Winter 2007), and one study over 12 months (Tremont 2008). In most studies, telephone counselling was performed weekly (Chang 1999; Davis 2004; Gant 2007; Glueckauf 2007; Steffen 2000; Winter 2007). The frequency of calls in one study was biweekly (Finkel 2007), and two studies started with weekly calls and then reduced stepwise after four or six weeks (Tremont 2008; Wilz unpublished). The duration of a single telephone counselling session also varied between studies. Sessions lasted either about 20 minutes (Chang 1999; Steffen 2000), approximately 30 minutes (Davis 2004; Finkel 2007), or 60 minutes (Wilz unpublished; Winter 2007). In one study, the sessions in the first half of the study lasted for 45 minutes and for 60 minutes in the second half of the study (Glueckauf 2007). Another study conducted a 60-minute session at the beginning and reduced the duration of subsequent sessions to 15 to 30 minutes (Tremont 2008). Studies with regular face-to-face counselling were excluded. However, single face-to-face contacts, for example at the beginning of the telephone counselling, were accepted. In six studies, counselling was only conducted via the telephone (Chang 1999; Gant 2007; Glueckauf 2007; Steffen 2000; Tremont 2008; Winter 2007). Two studies conducted an initial home visit (Davis 2004; Wilz unpublished), and one study conducted an additional athome session at the end (Finkel 2007). Study results can be influenced by the type of control group. It must be assumed that a treatment that is compared to usual care might more easily seem significant than a treatment that is compared to an active control treatment such as friendly calls. The study investigators of two trials stated that the participants in the control groups also judged the friendly calls as helpful (Chang 1999; Davis 2004). The differences in outcomes might therefore decrease, and it will become more difficult to find significance. Five studies compared telephone counselling with usual care (Glueckauf 2007; Steffen 2000; Tremont 2008; Wilz unpublished; Winter 2007), and one study treated the control group with usual care but added two brief telephone contacts at three and five months (Finkel 2007). Three studies used friendly calls in their control group (Chang 1999; Davis 2004; Gant 2007). Study participants The mean age of the participants in most trials lay between 60 and 66 years (Chang 1999; Finkel 2007; Glueckauf 2007; Steffen 2000; Tremont 2008; Wilz unpublished; Winter 2007). Only in Gant 2007 was the participants mean age 71 years, and the figures 15

18 stated in Davis 2004 seem to be faulty. Gender ratios differed between trials: two studies included only women (Chang 1999; Winter 2007), and one study included only men (Gant 2007). In three studies, the majority of participants were women (Finkel 2007; Glueckauf 2007; Wilz unpublished), and in one study the majority were men (Steffen 2000). The numbers stated in Davis 2004 seem to be faulty in this respect also. Three trials did not report the duration for which the carers had been caring (Finkel 2007; Gant 2007; Steffen 2000), and this varied in the other included studies. Davis 2004 reported a mean length of caring of approximately two years; Chang 1999 and Tremont 2008, a mean of about three years; Winter 2007, a minimum of about four years and Glueckauf 2007 and Wilz unpublished, a mean of more than five years. The relationship between the carer and the person with dementia also differed between studies. In five studies, more than half of the participants were caring for their spouses (Chang 1999; Gant 2007; Steffen 2000; Tremont 2008; Winter 2007), but in three other studies, less than half were caring for their spouses (Davis 2004; Finkel 2007; Glueckauf 2007). In Wilz unpublished, 45% of the carers were of the same generation as the person they were caring for and 49% were of a younger generation. Six other studies also included participants who were caring for a parent (Davis 2004; Finkel 2007; Gant 2007; Glueckauf 2007; Steffen 2000; Tremont 2008); in four studies, participants also included other family members - not only the children of the person with dementia (Davis 2004; Gant 2007; Glueckauf 2007; Steffen 2000). Two studies only stated how many of the carers were spouses, but did not identify the family relationship of the remaining participants (Chang 1999; Winter 2007). Some studies excluded carers who reported alcohol abuse problems (Davis 2004; Gant 2007; Steffen 2000), suicidal ideation (Davis 2004; Gant 2007; Glueckauf 2007; Steffen 2000), or other psychological difficulties (Glueckauf 2007; Steffen 2000; Tremont 2008). Measurement of outcomes All the included quantitative trials reported their outcome measures as mean differences with standard deviations (SDs). However, Glueckauf 2007 reported results as medians with semi-quartile, and inter-quartile ranges, and Finkel 2007 reported t-statistics instead of standard deviations (SDs). The conversion from median into mean differences is possible, but, despite a direct request, the study investigators did not supply the necessary minimum and maximum data. Therefore, it was not possible to include the results of Glueckauf 2007 in the meta-analysis; instead the results were described in narrative form and the t-values converted into SDs. Winter 2007 did not state the outcome results in combination with the number of participants in each group, and we therefore allocated the dropouts from this trial according to the distribution at baseline. Outcomes related to the carer Primary outcome: depressive symptoms Eight studies assessed depressive symptoms (Chang 1999; Davis 2004; Finkel 2007; Glueckauf 2007; Steffen 2000; Tremont 2008; Wilz unpublished; Winter 2007). Most used the Centers for Epidemiologic Studies Depression Scale (CES-D) (Finkel 2007; Glueckauf 2007; Wilz unpublished; Winter 2007), two trials used the Geriatric Depression Scale (GDS) (Davis 2004; Tremont 2008), one study used the Brief Symptom Inventory (BSI) (Chang 1999), and one study used the Beck Depression Inventory (BDI) (Steffen 2000). Secondary outcomes Burden Caregiver burden was assessed by seven studies using different tools. The Zarit Burden Interview was used by two studies (Tremont 2008; Winter 2007), and the Caregiver Appraisal Inventory (CAI) was used by two studies (Chang 1999; Glueckauf 2007). The Screen for Caregiver Burden (SCB) was used by Davis 2004, the Revised Memory and Behavior Problems Checklist (RMBPC) by Finkel 2007, and the Subunits Upset and Annoyance from RMBPC by Gant Distress Carer distress was assessed by two studies (Davis 2004; Gant 2007). Davis 2004 used scale B of the Screen Caregiver Burden to measure it, while Gant 2007 used the Target Complaints Interview Assessment Tool. Anxiety Only Chang 1999 assessed carer anxiety, using the Brief Symptom Inventory (BSI). Quality of life Tremont 2008 was the only study that assessed quality of life, using the quality of life parameter of the SF36 General Health tool. 16

19 Self-efficacy Four studies assessed carer self-efficacy (Gant 2007; Glueckauf 2007; Steffen 2000; Tremont 2008). Three used different versions of the Caregiving Self-Efficacy Scale by Steffen (Gant 2007; Glueckauf 2007; Steffen 2000), while Tremont 2008 used the Self- Efficacy Scale by Fortinsky. Satisfaction Two studies assessed carer satisfaction using different tools: Chang 1999 used the Caregiver Appraisal tool, and Davis 2004 used the Caregiver life satisfaction (LSI-Z) measure. Social support Carers social support was assessed by three studies with different measures (Davis 2004; Finkel 2007; Tremont 2008). Davis 2004 assessed the social support with the Caregiver Social Support tool (ISEL); Finkel 2007 used a social support scale from the Inventory of Socially Supportive Behaviors; and Tremont 2008 used the Multidimensional Scale of Perceived Social Support (MSPSS). Outcomes related to the care recipient None of the included quantitative trials assessed outcomes such as institutionalisation, mood or quality of life of the care recipients. Five of the quantitative trials reported how many of the care recipients were institutionalised as reasons for loss to follow-up (Davis 2004; Finkel 2007; Gant 2007; Glueckauf 2007; Winter 2007). Time of assessment of outcome measurements The included studies assessed outcomes at different time points. The outcomes of studies that conducted telephone counselling without an additional intervention were assessed after three months (Davis 2004; Glueckauf 2007; Wilz unpublished), six months (Davis 2004; Finkel 2007; Winter 2007), nine months (Wilz unpublished), and after 12 months (Tremont 2008). The outcomes of the study that conducted telephone counselling in combination with video sessions were assessed after one, two, and three months (Chang 1999). The outcomes of the studies that conducted telephone counselling in combination with video sessions and a workbook were assessed after two months (Steffen 2000), and after three months (Gant 2007). Funding source All except one of the included quantitative trials reported their funding source (Steffen 2000). Four trials were supported by an Alzheimer s organisation (Chang 1999; Gant 2007; Glueckauf 2007; Winter 2007), two received private funding (Finkel 2007; Glueckauf 2007), and four received governmental funding (Davis 2004; Gant 2007; Tremont 2008; Wilz unpublished). Evaluation of experience The thematic synthesis included two descriptive studies. Chang 2004 was a qualitative process evaluation of a larger RCT that could not be identified. The review authors contacted the trial authors but received no response. The larger trial, which was partly described within Chang 2004, used the same methods and the same population as Chang 1999, however, these studies were not the same. The Salfi study was first published as a doctoral thesis in 2004 and in 2005 was published as an article (Salfi 2005). We compared and summarised the findings presented in both documents; the study of Salfi 2005, was standalone qualitative research. Study respondents Both qualitative studies explored the perspective of the carers of people with dementia who received telephone counselling (Chang 2004; Salfi 2005). The Salfi 2005 trial aimed to understand the characteristics of telephone counselling comprehensively and to learn more about how both the carers and the counsellors experienced it, so observed not only the carers, but also the counsellors. The process evaluation in Chang 2004 involved the counsellor recording notes in real time during semi-structured telephone delivery of the cognitive behavioural therapy (CBT)-based intervention to carers. These handwritten notes were later subjected to content analysis. The process evaluation involved 83 participants in an RCT of an intervention following the same approach described in Chang Forty-one of the participants were assigned to the intervention group and 42 to the control group: 75 of the 83 participants were female, and the mean age of the whole study population was 68 years. Sixty-seven participants were white, 12 were black, and four were Hispanic. On average, all participants went to school for 14 years, and 47% of them earned between USD 10,000 and USD 20,000. They had been supporting people with dementia for approximately 20 hours a day for an average of 3.2 years. In the Salfi 2005 study eight carers and four counsellors were interviewed. Six of the carers were between 50 and 59 years old, two were between 60 and 80 years old, and all of them were female. Three of the eight were taking care of their husbands while the other five were in charge of their fathers. One carer had been nursing for less than a year and another one for one to two years; four carers had been nursing for between three and five years, and two of them had been in charge of a person with dementia for more than five years. The telephone counsellors were aged between 30 and 65 years. They had between three and 15 years experience counselling, two had a degree in nursing, and two had a Bachelor of Arts degree in gerontology. 17

20 Setting Both studies were conducted in high-income countries (World Bank 2013). One study was based in Canada (Salfi 2005), and one in the USA (Chang 2004). The Chang 2004 study included community-based participants, but the Salfi 2005 study did not state clearly if it took place in an urban or rural setting. The telephone counselling in Chang 2004 was performed within an RCT intervention. The telephone counselling in Salfi 2005 was established support offered by two agencies and not part of any evaluation. One agency was a branch of the Alzheimer Society and the other a multiservice community agency. It was not clear whether both agencies were non-governmental organisations. Telephone counselling The telephone counselling in both studies was poorly described. The available information is stated below. Participants in Chang 2004 who were assigned to the intervention group were visited at the beginning of the study and provided with a video with information about dressing and feeding people with dementia. After that, semi-structured telephone counselling was performed every two weeks for a total of 12 weeks. During each counselling session, existing problems and problem-solving strategies were discussed and adapted to each individual case. In addition to dressing and feeding people with dementia, other topics addressed by the carers were discussed according to the problemsolving guidelines. The telephone counselling sessions lasted from five to 90 minutes, with an average of 18 minutes, and were performed by two counsellors with Masters degrees and a gerontological nursing specialisation. The hand-written notes of the counsellors from these counselling sessions constituted the raw data for analysis. Two different counselling institutions attended to the participants in the Salfi 2005 study. The first institution offered counselling for eight hours daily on work days only. The second institution also offered a daily program for adults involved, and training courses in addition to telephone counselling. Carers had access to all of these services and were not limited to telephone counselling. Both institutions offered information and emotional support during their counselling. Funding source Both studies stated their source of funding: both were funded by an Alzheimer s organisation, and the Salfi 2005 study received additional financial support from a community care centre. Excluded studies Trials were excluded if they did not fulfil the inclusion criteria. Most of the 98 excluded studies were clearly outside the scope. Four studies are listed, with the reasons for their exclusion, in Characteristics of excluded studies since their exclusion was not clear at first glance. Risk of bias in included studies Evaluation of efficacy The results of the Risk of bias assessment are displayed in the Risk of bias graph and the Risk of bias table (see Figure 3 and Figure 4). Figure 3. Risk of bias graph: review authors judgements about each risk of bias item presented as percentages across all included studies 18