A cross-sectional, comparative study based on a conceptual model (Figure 1) was conducted among family caregivers and their non-caregiver counterparts. While each individual’s health is affected by factors such as demographics, health behaviors, and his or her own physical or mental health conditions, we also assumed that caregiving has a significant association on the health of the caregiver. Therefore, we compared health outcome data between family caregivers and non-caregivers counterparts, while adjusting for a variety of other possible factors affecting health outcomes.
Participants
“Caregiver” was defined as a main person taking care of an elderly family member who uses any home service under the LTCI system. Family caregivers were recruited into this study through 26 agencies providing LTCI services, such as care management, home care nursing or home help, in the urban areas of Tokyo, Osaka, Kobe, and Ibaraki.
We also recruited non-caregivers who were not providing care at the time of the survey. Non-caregivers were recruited through a university health checkup center located in an urban area in Tokyo. The center provided preventive health examinations; it did not provide any medical treatment, but could refer patients directly to the university hospital if any abnormal results were found. In both groups, additional inclusion criteria were the absence of any cognitive disorders and the ability to read and write Japanese without assistance.
Data collection procedure
We collected data on caregivers and non-caregivers using different procedures. Caregiver data were collected in participants’ homes by surveyors with nursing licenses, because caregivers have minimal time to spare to travel to a study site. First, the surveyors called potential participants to confirm their intention to participate in the study. During their visit, surveyors administered a questionnaire survey, then took a blood pressure reading and collected a blood sample using a kit [13]. We trained 13 registered nurses in a half-day training session, teaching them the rules and practical skills needed for the survey, as well as the standardized procedure for conducting blood pressure measurements and blood sampling using the kit. Data from non-caregivers were collected at the health checkup center. Non-caregiver questionnaires were self-administered because of time and space limitations at the center. Blood pressure readings and serum chemistry data were collected from the health checkup data.
In both groups, prior to the data collection, the surveyors explained to participants their rights to participate in or leave the study, the confidentiality of personal identification, and the freedom to leave the study without any disadvantage. After the potential participants agreed to participate in the study and signed a consent form, the data collection procedure was initiated. This study was approved by the ethics committee of Jikei University. The data were collected from June 2011 to August 2012.
Measurements
Questionnaires
The questionnaires consisted of questions regarding sociodemographic variables, physical condition variables, health-related behavior variables, and subjectively assessed health condition variables. Sociodemographic variables included age, sex, educational history, working status, household income, and marital status. Physical condition variables included the locomotive syndrome as measured by the Japanese Orthopedic Association method [14], weight, height, body mass index (BMI), and menopause status (female only). Locomotive syndrome was defined as lowered activities of daily living caused by disabilities of the locomotive organs and the risk of physical fragility [15]. The syndrome was assessed by five symptoms including pain, bone malformation, limitation of joint motion range, muscle loss, and loss of balance. Health-related behavior variables included alcohol consumption, smoking, and exercise. Subjectively assessed health condition variables included sleep disorder as measured by Pittsburgh Sleep Quality Index [16,17], depression as measured by the K6 [18], pain as measured by the Margolis rating system [19], and other health-related complaints [6] with history of medical treatments and medications. Information on care recipients and the caregiving situation was collected from caregivers, including sex, age, activities of daily living (ADL), instrumental activities of daily living (IADL), LTCI-certified care level, average time spent caregiving per day, or number of years of caregiving. The IADL was assessed by an instrument developed by Lawton [20]; it consists of 31 items in eight domains, such as making a phone call or going shopping. As an indicator of service use, we collected data on the amount of copayment spent for LTC services. The copayment for service use was 10% of the total cost under the LTCI.
Serum chemistry
The serum chemistry data used in this study were as follows: low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), aspartate aminotransferase (AST), alanine aminotransferase (ALT), gamma-glutamyl transpeptidase (gamma-GTP), uric acid (UA), creatinine (Cr) and glycosylated haemoglobin (HbA1c). Each variable was recoded into a dichotomous variable indicating whether the participant was within or outside of the normal range, based on the criteria described below.
Criteria for normalcy
High blood pressure
The criterion for high blood pressure was a blood pressure of 140 mmHg and more systolic (SBP) or 90 and more diastolic (DBP), based on the guidelines of the Japanese Society of Hypertension [21]. We also included individuals taking anti-hypertensive medications (receiving medication) in the high blood pressure group.
eGFR
The estimated glomerular filtration rate (eGFR) is a test to evaluate chronic kidney disease (CKD), often used in clinical practice or in epidemiological studies. Lower eGFR is considered a risk factor for CKD [22]. In this study, eGFR results calculated by standard formula for Japanese [22] were dichotomized into either less than 60 mL/min/1.73 m2 or within the normal range. Any participant with less than 60 mL eGFR was instructed to consult a physician [22].
Dyslipidemia
Standard primary prevention criteria for dyslipidemia were used: less than 40 mg/dL HDL-C or 120 mg/dL and more LDL-C [23]. We also included participants being treated for dyslipidemia in the “dyslipidemia” group.
Liver function
The criteria for liver function were as follows: more than 30 IU/L AST, more than 30 IU/LALT, or more than 50 IU/L γ-GTP, as recommended by the Japan Society of Ningen Dock [24]. We also categorized participants being treated for liver function in the out-of-normal-range group, with the exception of viral hepatitis. We included those with viral hepatitis as normal because their lowered liver function was not due to caregiving responsibilities.
Hyperuricemia
The criteria for hyperuricemia were either a UA level of 7.0 mg/dL and more, as suggested by the Japanese Society of Gout and Nucleic Acid Metabolism [25], or the current treatment of hyperuricemia.
Diabetes
The criteria for diabetes were either an HbA1c level of 6.5% and more according to the Japan Diabetes Society [26], or the current treatment of diabetes.
Data analyses
Since there was considerable difference between the caregiver and non-caregiver groups in terms of age and sex, we matched caregiver data to non-caregiver data based on these two characteristics for the purposes of the analyses, stratifying age into five year categories. We analyzed the male and female data separately, because previous studies have suggested biological and social sex differences in caregivers’ attitudes and the burden of caregiving, as well as the risk of the outcome variables [27].
Because the blood samples of the groups were obtained using different methods, we transformed the data in the caregivers’ sample using regression equations to adjust for systematic errors [13,28] and improve measurement reliability. Those regression equations were developed from separate blood samples from 48 persons obtained using both methods. From the data, we developed the equation y=b + ax (with y as data with the intravenous method used for non-caregivers and x as data with the kit method used for caregivers), and obtained estimates for “a” and “b.” The data of the caregivers’ blood samples (x) was thus transformed and the values obtained as y were used.
We first conducted bivariate analyses to examine the differences in sociodemographic, physical condition, health-related behavior and subjectively assessed health condition variables between caregivers and non-caregivers. Chi-square or Mann-Whitney U tests were used, depending on the type of data. Then, we conducted multivariate conditional logistic regression, using the dichotomous variables based on results of blood pressure and serum chemistry as outcome variables. In the regression model, in addition to the caregiver or not variable as an explained variable, we added the following control variables: education, household income, work, marital status, history of high blood pressure (eGFR model only), history of locomotive syndrome, BMI, menopause (only among females), smoking, alcohol consumption, exercise, sleep disorder, depression, pain, and indefinite complaints. All statistical analyses were performed using SPSS version 19.0 (Chicago, Illinois, IBM, USA).