The Dominican Republic shares the Caribbean island of Hispaniola with Haiti. The population is 9.4 million, and 0.5 million (5.7%) are aged 65 and over . Life expectancy is 71 years for men and 75 for women. The Dominican Republic is one of the poorest of the 10/66 Latin American countries, with a per capita GDP (PPP) of US$ 9,200. In common with other countries in the region there are high levels of income inequality (a Gini index of 51.6). Forty two percent of the population live below the poverty line, one third of these in extreme poverty. Pension coverage, at only 18% of the economically active population, is one of the lowest in Latin America. Extensive reforms have been proposed, and are in the process of implementation. Community health care is provided by the government through the system of 'primary attention units'. Consultations are free, but medicines must be paid for. Despite low medical insurance coverage, private healthcare is widely used instead. The Dominican Republic has only twenty psychiatrists, twenty psychologists, and two neurologists per million population.
Study design and catchment area: A one-phase cross-sectional whole population catchment area survey in geographically defined districts in Santo Domingo. The survey protocol is described in detail in an open access publication . Precision calculations indicated that an overall sample of 2,000 would allow estimation of a typical dementia prevalence of 2.5% with a precision of ± 0.9%. Atypical middle-class or high-income areas were avoided. The catchment areas selected were Villa Francisca, San Carlos, San Antón, Mejoramiento Social and Santa Barbara. After defining the boundaries, mapping was carried out to identify and locate households. Possible participants (inclusion criterion age 65 and over) were identified. Age was formally determined on revisit for the interview by comparing stated age according to participant and informant, and official records if available. Irreconcilable discrepancies of three or more years were settled using an event calendar approach. Participants were recruited following informed consent or on the basis of a relative's agreement in case of lack of capacity for consent due to dementia. Interviews were carried out in participants' own homes. All participants received the full assessment lasting approximately two to three hours. Ethical approval for the survey was provided by the research ethics committee for the Institute of Psychiatry, King's College London, and the Bioethics National Committee for Research in the Dominican Republic.
1) Outcome – dependency
The interviewer used a series of open ended questions in their interview with a key informant, to establish the presence or absence of dependency:
a) Who shares the home with the participant?
b) What kind of help does the participant need – inside of the home? – outside of the home?
c) Who, in the family, is available to care for the participant?
d) What help do you provide?
e) Do you help to organise care and support for the participant?
f) Is there anyone else in the family who is more involved in helping than you? What do they do? What about friends and neighbours? What do they do?
On the basis of the answers to these questions, the interviewer defined the family network, established if the older person needed and got any care from family members or others, identified who was responsible for organising and providing 'hands on' care, and if the informant was one of the main caregivers. Finally, they coded whether the older person required a) no care, b) care, some of the time or c) care, much of the time.
2) Health conditions contributing to dependency
a) Dementia according to either the 10/66 dementia diagnosis algorithm  or DSM IV dementia criterion .
b) Physical illness. Self-reported stroke, and having three or more of 11 limiting physical impairments (arthritis or rheumatism; eyesight problems; hearing difficulty or deafness; persistent cough; breathlessness, difficulty breathing or asthma; high blood pressure; heart trouble or angina; stomach or intestine problems; faints or blackouts; paralysis, weakness or loss of one leg or arm; skin disorders such as pressure sores, leg ulcers or severe burns) 
c) Depression – ICD 10 depressive episode (mild, moderate or severe), derived using a computerised algorithm applied to a structured clinical interview, the Geriatric Mental State .
3) Characterisation of those with dependency
a) Disability Activity limitation and participation restriction measured by the WHO-DAS II , specifically developed by the WHO as a culture-fair assessment tool for use in cross-cultural comparative epidemiological and health services research.
b) Care arrangements (only assessed among those needing care)
Time spent by the caregiver in the last 24 hours in specific caregiving activities ; communicating, using transport, dressing, eating, looking after one's appearance, and supervising,
c) Impact of providing care on caregivers (only assessed among those needing care)
Caregiver perceived strain – the Zarit Burden Interview [18–20] with 22 items that assess the caregiver's appraisal of the impact their involvement has had on their lives.
Economic strain – the extent to which the caregiver had cut back on or stopped work in order to provide care, and paid care inputs
d) Caregiver mental health – the Self Reporting Questionnaire 20 [21, 22].
e) Behavioural and Psychological symptoms of dementia; the Neuropsychiatric Inventory, NPI-Q .
The assessments of care arrangements and caregiver strain were applied and refined in the previous 10/66 Dementia Research Group pilot studies, conducted in 26 LAMIC centres, including the Dominican Republic. The Zarit Burden Interview was found to have robust cross-cultural measurement properties [9, 24], but has not been specifically validated for this population. The NPI-Q has been previously validated in Spanish , but we are not aware of any previous use in the Dominican Republic.
1) We report the prevalence of dependency (needing some care, needing much care, needing any care) by age and gender using Stata 9.2 survey commands to generate robust standard errors and 95% confidence intervals, taking account of household clustering.
2) We describe the sociodemographic characteristics, economic circumstances and health status of the sample by levels of dependency. We used a Poisson regression model (adjusted for household clustering) to estimate the independent associations of dementia, three or more physical impairments, stroke and ICD-10 depression with needing much care, controlling for age, gender, education and marital status. The resulting prevalence ratios, together with the prevalence of the exposure were used (STATA aflogit command) to calculate population attributable prevalence fractions (PAPFs) as an index of the salience of each health condition to the prevalence of dependency at the population level.
3) Among those needing care, we compare the health characteristics, care inputs, and indicators of caregiver strain between those with dementia (and their caregivers) and those with other health conditions (and their caregivers).
4) We generated a general linear model to estimate the independent contributions of dementia, depression, physical impairments and stroke to caregiver strain measured using the Zarit Burden Interview, controlling for the main sociodemographic characteristics of the care recipient and the caregiver, and their relationship. The proportion of the variance accounted for by these factors was estimated before and after adding, separately, time spent assisting with activities of daily living, and the severity of behavioural and psychological symptoms in the care recipient as potential mediating variables in the association between dementia and caregiver strain.