The study population consisted of caregivers of elderly patients with AD (n=25) and patients admitted to the Association for Research and Assistance to Patients with Alzheimer's (AEPAPA) (n=25) in the city of Guarapuava (PR). In order to determine the size of the sample, we considered all patients that fit the inclusion criteria of the study and are serviced by AEPAPA. In this study, the age cut-off used to classify patients as elderly was based on the criteria proposed by the World Health Organization (WHO), which defines individuals aged 60 years or older from developing countries as elderly. In order to identify and select the caregivers, AEPAPA was asked for an official list, containing identification, address and telephone number of the registered elderly. Subsequently, the cadastral data was confirmed with the Social Worker, with the intention of eliminating the names of the deceased or who had changed their residence or changed their city. A telephone appointment was initially made by the social worker of the AEPAPA. In case of refusal or absence (three attempts were made to contact each domicile), the subject was excluded. On the day of the visit, the professional, together with a nurse from the AEPAPA, made the verbal invitation to participate in the research, by means of a previous explanation of all the stages of the study and, later completing the Informed Consent Form (TCLE) by the participants. In the case of illiterate caregivers, the informed consent form was signed by a family member responsible after reading the terms to the caregiver. AEPAPA is a Civil Society with a care purpose, responsible for providing home care to the elderly, with Public Utility, according to Law No. 2157/2013, contained in the Official Gazette of the Municipality from August 24 to 30, 2013. The ethical precepts of voluntary, enlightened and consensual participation of each participant were respected through a Free and Informed Consent Term – informed consent form signed by the participant (caregiver and patient). The project was approved by the Ethics Committee on Human Research of the Health Sciences Sector, the State University of West Paraná Center (896 296/2014).

This descriptive cross-sectional study began with the collection of data and salivary material from caregivers in the home of the elderly with AD, by a team composed of a nurse, a psychologist, a social worker, a pharmacist and a nutritionist from AEPAPA in the period from December 2014 to December 2015. The study was quantitative and used formal instruments to collect the data as questionnaires, scales and tests based on the work of 25. Initially, the caregivers who agreed to participate were asked to refrain from eating, drinking caffeine, brushing their teeth and vigorous exercise for 2 h before the collection of cortisol.

Participants provided the first saliva sample in the morning, between 08:00 and 09:00 hours, after which the questionnaires were applied to the caregiver and the elderly patient. Initially, the sociodemographic characterization of the caregivers was based on the following aspects: relationship to the elderly patient, age, race, marital status, number of children, schooling, income, type of housing, hygiene conditions and basic sanitation in the residence where they performed their care work. The remaining tests included information such as the analysis of the caregiver’s quality of life through the 36-Item Short Form Survey (SF-36) and the investigation of affective and somatic cognitive behavioral manifestations in the face of a possible picture of depression presented by the caregiver with the application of the Beck Depression Inventory (BDI), as well as the evaluation of the Instrumental Activities of Daily Living of the patient (IADL) with the help of the Lawton Scale, which focuses on assessing the elderly person's ability to maintain an independent life that interferes with the degree of caregiver activity, and classification of the stage of Alzheimer's disease using the Clinical Dementia Rating (CDR) ^{26– 29} .

The other saliva samples were collected in the afternoon, between 16:00 and 17:00 hours, and at night, between 22:00 and 23:00 hours, taking into account the circadian fluctuations of cortisol. After collection in a Salivette tube (Salivette®, Sarstedt, Alemanha), since the biochemical dosage was performed on the saliva, the samples were immediately stored in a test tube (typically Ultra-High Performance 15 ml centrifuge tubes; VWR, Radnor, PA). At the end of the sample collection, the samples were sent to the Master Laboratory, a private clinical laboratory, in Guarapuava, PR, with no charge on the patient.

In the laboratory, the samples were refrigerated in a freezer (-4°C) for 24 hours until processing. Extraction of Salivette cotton saliva was performed by centrifugation (10 min at 3000 rpm) being the cortisol dosage in the saliva was performed by electrochemiluminescence immunoassay according to the instructions of the manufacturer (ECLIA; catalog number 11875116160; Roche Diagnosis GmbH, Mannheim, Germany).

The SF-36 questionnaire, applied to caregivers in the study, has the advantage of being easy to understand and requiring a short application time; therefore, it is well suited to assessing the associations of several types of diseases with the quality of life of those involved. SF-36 consists of 36 general questions grouped into eight domains: functional capacity (10 items), vitality ⁴ , physical ⁴ , pain ² , general health ⁵ , social ² , emotional ³ and mental health ⁵ . Each of these questions is answered with a value between 0 and 100, where 0 is the worst state of health and 100 is the perfect state of health ²⁶ .

The BDI, applied to caregivers in the study, is an instrument structured in 21 descriptions of affective and somatic cognitive symptoms of depression, with answers to each question ranging from 0 to 3 points dealing with the absence of depressive symptoms until the presence of symptoms. Originally created by Aaron Temkin Beck ³⁰ , satisfactory evidence of its trustworthiness and validity was shown by a study by Cunha, 2001 ³¹ . To get a total score to be evaluated, the points of each item are added, giving a total score whose maximum does not exceed 63 points. According to Caixeta, scores from 0 to 9 points indicate an absence of depression or minimal depressive symptoms; from 10 to 18 points indicate mild to moderate depression, from 19 to 29 points indicate moderate to severe depression, and 30 to 63 points indicate severe depression ²⁸ .

The evaluation of the implementation capacity of IADL was performed by applying the Lawton Scale to patients with AD in the study. For each task, there are three possible answers, with scores ranging from 1 to 3 (1, dependence; 2, ability to perform the task with help; and 3, independence). The final score is reached by the sum of points of the eight domains and ranges from 8 to 24; the higher the score, the more independent the individual ^{27, 32} .

The AD CDR scale, applied to elderly patients in the study, was developed in 1979 at University of Washington's, St. Louis, Missouri, through the project" Memory and Aging" ³³ , to graduate dementia and classify patients according to the disease stage, evaluating the presence or absence of cognitive impairment on a five-point scale (0, 0.5, 1, 2, 3). The six domains analyzed by the test were: memory, orientation, judgment and problem solving, community function, home and hobby function, and personal care. A general CDR score was then performed using the individual ratings in the 6 areas, according to the standard scoring rules ³⁴ ; so that a CDR score of 0 indicates no dementia, while CDR scores of 0.5 1, 2 and 3 indicate very mild, mild, moderate and severe dementia, respectively. In addition, individual domain scores can be summed, totaling a score ranging from 0 (0×6: when there is no impairment in any domain) to 18 (3×6: when there is maximum commitment in all domains) ²⁹ .

To describe the results, descriptive tests were carried out with measures of central tendency and frequencies. To verify normality of the sample, the Shapiro–Wilk test was calculated. Pearson’s and Spearman’s correlation analyses were applied to parametric and non-parametric data, respectively. Also, for the association between CDR and cortisol values, the Kruskal–Wallis test was applied. The level of significance was set at 5% (P<0.05). The analyses were performed using the Statistical Package for the Social Sciences software version 22.0.

Of the total sample, 88.0% (n=22) of the caregivers were female, 68.0% (n=17) of whom were Caucasian. Regarding marital status, 56.0% of participants (n=14) were married and 84.0% (n=21) had up to 4 children, while 16.0% (n=4) reported having 5 to 8 children. Regarding the level of education, 8.0% of caregivers (n=2) were illiterate, 44.0% (n=11) did not complete elementary school and only 4.0% (n=1) had completed higher education. Regarding the type of housing, 92.0% (n=23) resided in brick houses, while 8% (n=2) reported living in wooden houses. 76.0% (n=19) rated their hygiene conditions as good, compared to 24% (n=6) who reported very good hygiene conditions; all caregivers had water, electricity and sewage in their homes. As to income, 88.0% (n=22) of caregivers reported monthly income less than two minimum wages, equivalent to US $583.00. By contrast, only 12% (n=3) reported an income of up to three minimum wages.

The distribution of participants in terms of their relationship to the patients was predominantly close to the patients, with 44.0% (n=11) and 24.0% (n=6) of caregivers self-described wives or husbands, respectively. Caregivers were family members in 80.0% (n=20) of cases; among these, 24.0% (n=6) of caregivers self-declared as wives or husbands, 8.0% (n=2) as grandchildren and 4.0% (n=1) nephews. The remaining 20.0% (n=5) were unrelated caregivers.

With respect to depression affecting caregivers, values obtained from the BDI resulted in a mean score of 15.68 (SD=10.05), indicating mild to moderate depression in 36.0% (n=9) of the caregivers, while 20.0% (n=5) of caregivers were classified as having moderate to severe depression, 12.0% (n=3) as having severe depression and 32.0% (n=8) as not having depression.

SF-36 variables showed close variations between functional capacity and social aspects among caregivers, with a mean score of 74.2 (SD=28.7) for the functional capacity and 64.4 (SD=28.1) for the social aspect. Regarding functional capacity, the variable with the lowest mean score was the physical aspect (mean=50, SD=46.8), compared to the aspects of pain (mean=53, SD=22.9), general health status (mean=52.7, SD=15.6) and vitality (mean=53.2, SD=19). As for the social aspects, the emotional variable (mean=40, SD=48.1) had the lowest score compared with mental health (mean=54.2, SD=21.7).

In the evaluation of IADL, the results showed that 70.83% (n=17) of patients with AD were not able to maintain an independent life, and 20.83% (n = 5) were unable to perform any of the nine corresponding functions of the scale. In total, 50.0% (n=12) performed some activity with partial help, and 29.17% (n=7) were able to perform the activities without help. One patient did not participate in the questionnaire.

Regarding the stage of the disease 16.0% (n=4) were in the mild stage of AD (CDR 1), 52.0% (n=13) in the moderate stage (CDR 2) and 32.0% (n=8) in the severe stage (CDR 3).

For the salivary cortisol test, 3 samples were collected from each patient, accounting for a total of 75 samples. With regard to the salivary cortisol dosage in the morning, the median value was 0.50 μg/dl, with a minimum of 0.33 μg/dl and maximum of 1.27 μg/dl (reference value, approximately 0.69 μg/dl). Salivary cortisol measured in the afternoon resulted in a median value of 0.32 μg/dl, with a minimum of 0.09 μg/dl and a maximum of 0.71 μg/dl (reference value, approximately 0.43 μg/dl). For the tests performed in the evening, the median value was 0.68 μg/dl, with a minimum of 0.32 μg/dl and a maximum of 1.26 μg/dl (reference value, approximately 0.35 μg / dl). The cortisol values were analyzed according to the values cited by Aardal and Holm, 1995 ³⁵ .

As analyzed in the paragraph above, the percentage of caregivers who presented cortisol at normal concentration was considerably higher than the percentage of caregivers who exhibited cortisol at an altered concentration, in the collection periods (morning, afternoon and night). In the morning, 80.0% (n = 20) of the caregivers did not present alterations in cortisol dosages, according to the reference values used ³⁵ , whereas in the afternoon the percentage dropped to 72.0% (n = 18) and 60.0% (n = 15) at night, according to the circadian cycle of cortisol, evidenced in healthy individuals.

Of the variables analyzed by the study, only the age of the caregivers was significantly correlated with the dose of salivary cortisol collected in the afternoon (r=0.470; P=0.018), as well as when it was correlated with daily cortisol variation, corresponding to mean of the cortisol dosage in the three periods (r=0.403; P=0.046) ( Table 1).

Figure 1, Figure 2 and Figure 3 show a positive correlation but no significant variation between nocturnal salivary cortisol (μg/dl) and the scores from AIVD, IDB and SF-36. Figure 4 shows a positive and significant correlation (r=0.403; P=0.046) between the daily variation of cortisol dosage (μg/dl) and the age of caregivers.

All correlations with significant p values were considered weak, owing to the sample size. However, it is possible to observe that cortisol values in the morning were correlated with the degree of kinship (r=0.289; p<0.05). The correlation was weak but inverse between the degree of kinship and the daily variation of cortisol (R²=-0.311; p<0.05).

Table 2 shows the relationship between the CDR scores of the patients with AD and the corresponding mean salivary cortisol doses (mg/dl) of their caregivers. Mean values of salivary cortisol of the caregivers, measured in the morning and afternoon periods, remained below the cortisol values considered normal for these periods, which are approximately 0.69 μg/dl and 0.43 μg/dl, respectively, for the three groups of investigated CDRs. This diverged from the mean cortisol scored at night, which remained above the reference value considered normal for this period, approximately 0.35 μg / dl, for the three analyzed groups.

In a horizontal analysis of the table, the mean cortisol levels in the specific periods evaluated (morning, afternoon and evening) and the mean daily cortisol variation did not change significantly (P>0.05) when associated with the CDR groups ( Table 2).