We will include studies dealing with primary informal caregivers of community-dwelling persons with AD. These unpaid caregivers are typically close relatives (for example, spouses, children) who provide the bulk of care and support. Studies including only paid caregivers, such as home help aides or employees of long-term care facilities, as well as studies set only in institutions, will be excluded from the review. There will be no restrictions on the sex, age, or ethnic background of caregivers.
Included studies must report at least one of the following outcomes in relation to caregiver QoL: level of care, quality of care, amount of time spent providing care, and similar variables dealing with caregiver performance in their caregiving role. We will include primary studies that provide quantitative results. We will include studies written in any language that are experimental (including RCTs, quasi-randomized trials, controlled clinical trials), quasi-experimental (including interrupted time series and controlled before and after studies), or observational designs with comparison groups.
Level or quality of care
Due to the difficulty involved in determining what constitutes ‘quality’ care, studies using any scale that was designed to measure quality will be considered for inclusion. This includes studies measuring ‘level’ or ‘quality’ of care from a task-oriented perspective.
The concept of QoL is multi-faceted and may be defined in several ways. In this review, QoL will be treated as an independent variable. The most common approach to measuring QoL involves the use of scales, several of which have been developed for healthcare research . Popular, generic, health-related QoL scales include the Short Form 36 (SF-36) , EuroQoL Group’s EQ-5D , and the World Health Organization Quality of Life-BREF (WHOQOL-BREF) . Other QoL scales, such as the Quality-of-life - Alzheimer’s Disease (QOL-AD) scale , were specifically developed for use in patient populations or diseases.
We will include any study measuring QoL, regardless of the specific scale used. In addition to QoL, studies examining the impact of the following constructs on our outcomes will be included.
Social support is a multidimensional construct of the extent to which individuals receive emotional support, instrumental assistance, information, guidance and feedback, personal appraisal support, and companionship from family members, friends, co-workers, other persons (for example, acquaintances, religious leaders, therapists), or organizations (for example, caregiver support groups).
Caregiver burden is operationalized by any construct representing the physical, emotional, and financial cost of providing care for a loved one with AD. The Zarit Burden Interview (ZBI) [8, 18] is a widely used instrument for measuring caregiver burden.
Although depression is an element of QoL, it is also an important construct on its own. Indeed, depression is one of the common side effects of long-term caregiving . Depression can be measured by several instruments including the Center for Epidemiologic Studies-Depression (CES-D) scale [20, 21].
We will include any study measuring these constructs, regardless of the means by which these constructs are measured.
Included studies may present quantitative results only.
Two reviewers will independently screen the titles and abstracts of studies identified in the literature search. Studies meeting the eligibility criteria, or studies whose titles and abstracts do not provide sufficient information to assess eligibility, will advance to full-text screening. During full-text screening, the reviewers will independently read entire papers and assess eligibility. Conflicts will be resolved by discussion between the two reviewers or by the involvement of a third reviewer.
Data collection process
A detailed data collection form will be developed to collect information about study characteristics (study design, sample size, year, country), participant characteristics (for example, type and number of patients, caregiver relationship to person with AD, living situation, age, mean and standard deviation, AD diagnosis criteria), and results (for example, quality of life, quality of care, institutionalization, amount of care, and so on).
Data will be extracted using an online form uploaded to DistillerSR. The data extraction form will be piloted by two reviewers and further refined if necessary. Two reviewers will independently extract all of the data to achieve the highest level of accuracy. Reviewers will meet to resolve discrepancies. If the two reviewers cannot agree, then a third reviewer will be asked to adjudicate.
In cases where studies report outcome results over different time periods, we will extract data from each time period to examine the impact of the intervention over time. Further, in cases where multiple publications report data from the same study, we will use the most current report of the outcomes of interest. When data are not clearly reported, we will contact the lead author of the study for clarification.
Assessment of risk of bias
Each included study will be assessed for risk of bias using appropriate quality assessment tools. For studies employing RCT design, we will use the Cochrane Risk of Bias Tool . The Cochrane Effective Practice and Organisation of Care Risk of Bias Tool will be used for assessment of risk of bias for controlled clinical trials, interrupted time series, and controlled before-after studies [23, 24]. Finally, the Newcastle-Ottawa Scale will be used for studies employing cohort and case control designs .
We will use GRADE to assess the level of evidence across studies and also use funnel plots to assess publication bias .
A narrative synthesis method will be used to describe the results. All included studies will be summarized in narrative form, and summary tables will be created showing key study characteristics (that is, population characteristics, treatment interventions, study outcomes, sample sizes, settings, funding sources, and comparator treatments (type, duration, and provider)), methodological limitations, and any other important aspect related to each research question of interest.
Meta-analysis will be performed if possible from the data extracted from the included studies. If clinical groups are too heterogeneous to permit meta-analysis, a separate qualitative analysis will be presented and graphical representation may be used to display main study outcomes.
If outcomes of interest in each included study were reported using different outcome measures on a continuous scale, the DerSimonian and Laird random effects models with inverse variance method will be utilized to generate the summary measures of effect in the form of standardized mean difference (SMD) for each outcome . The use of SMD as a summary statistic in a systematic review is appropriate if the same oucome (that is, quality of care) is assessed in a variety of ways or if different psychometric scales were used . In such a situation, it would be necessary to standardize the results of the studies before they could be compared across studies or combined in a quantitative synthesis.
Standardized mean differences (SMDs) will be calculated using change from baseline data, that is, mean difference between pre-treatment (baseline) and post-treatment (final/endpoint) scores along with its standard deviation for both intervention and control groups. Appropriate correlation between pre-treatment (baseline) and post-treatment (final/endpoint) level of care scores will be used based on evidence from existing literature. The Cochran’s Q (α=0.10) and I2 statistic will be utilized to quantify the statistical heterogeneity between studies examining our outcomes of interest, where P <0.10 indicates a high level of statistical heterogenity between these studies. Sensitivity analyses will be performed on the type of intervention, study risk of bias and by removing studies with obvious between-group baseline imbalance in order to evaluate statistical stability and effect on statistical heterogeneity.