Incontinence problems have been identified as a significant factor in increasing carer ‘burden’ and triggering the entry of people with dementias to care homes.110,134,356 Supporting people with dementia to live at home is a major policy objective. There are currently no primary care-based data available in the UK or elsewhere to provide clinicians or commissioners services with data on the incidence and current management of people with dementia living at home who also have problems with urinary and/or faecal incontinence. This protocol addresses this absence of data through an analysis of general practice records held in the THIN database.
Background
Dementia affects over 4% of the people over 60, increasing to 13% for those over the age of 80 in Western Europe with significant projected increases in the next 20 years.345 Dementia has enormous impact not only for the individual and their family but also for the health and social care system.119,296,346 The median length of time from diagnosis to death is 10 years for those under 65 at diagnosis and 4 years for those over 80 years.173 The clinical syndrome of dementia has a trajectory of progressive deterioration in cognition, abilities, and physical functioning. The impairment experienced by the individual is often compounded by extrinsic factors such as attitudes of ‘therapeutic nihilism’115 in professionals, un-adapted environments, and social exclusion. As the condition progresses problems may arise in maintaining independence in going to the toilet, in avoiding constipation, and in managing incontinence problems that cannot be resolved. Estimates suggest that there are 500,000 people with dementia in England and two-thirds live at home.119
Incontinence is the involuntary leakage of urine or stool or both.347,349 The prevalence of any type of urinary incontinence in all older adults is between 6% and 10%, with increasing rates associated with old age.349 It is estimated that 2–5% of adults experience faecal incontinence. Incontinence lowers quality of life and impacts negatively on mental health351,352 as well as creates significant practical and financial problems. There is evidence that incontinence contributes significantly to other major health issues for older people such as falls.353 Incontinence has been identified by professionals and older people as an issue of unmet need in primary care.354,355,381
There are no population surveys that identify the incidence of people with dementia and continence problems living at home. However, there are some indications that they may be a substantial minority. A UK population-based study of 15,000 home-dwelling people aged over 75 identified that 18.3% had cognitive impairment and of these 31% had urinary incontinence problems.135 A national general practice audit of 999 older patients with faecal incontinence identified that 27% had a diagnosis of dementia.367 There are some indications that people with dementia and incontinence symptoms are not investigated and managed proactively.407
There are currently no primary care-based data available in the UK or elsewhere to provide clinicians or commissioners services with data on incidence and the characteristics of people with dementia living at home who also have problems with urinary and/or faecal incontinence.
The experience of people with dementia and their family carers interviewed in a related ongoing study by the authors suggests that:
Like the general population
408 most delay consulting general practice or any health professional about incontinence problems as they feel embarrassed and humiliated.
Family members report avoiding talking to their general practitioner about these issues as they seek to protect the person with dementia’s dignity and public persona.
354 However, as the dementia progresses this becomes unsustainable and the family carers insist on seeking help in order to cope with the wider burden of supporting their family member. This is sometimes reported as part of a crisis situation in which hospitalisation or temporary residence in a care home occurs.
Family carers report a variety of responses from general practice when they do seek help: some indicate they get little assessment or help other than the referral for continence pads, others indicate a more detailed investigative approach to identify or discount possible treatable causes of these symptoms, such as urinary tract infections or prostate hypertrophy problems, and then more active management of the problems.
Purpose
This cohort study will describe current practice in continence care for people with dementia, and quantify the burden of disease. It will also examine whether this differs with demographics or health status, or between general practices in ways not explained by patient characteristics.
Hypotheses
The background described above suggests a number of hypotheses:
The general practice reported incidence of incontinence problems is higher in the dementia population than in the general population of similar age registered with a GP. This relationship may be confounded by other factors such as age or mobility.
The general practice response to people with dementia consulting for incontinence problems, compared to people of similar demographics without dementia, is more likely to be prescription of continence pads, and less likely to be investigation and referral for tractable problems, including surgery, or it may take longer to get referred.
Research questions
What is the incidence rate of urinary and/or faecal incontinence, stratified by age and sex, recorded in general practice patients aged over 60:
Which covariates affect this relationship (co-morbidity, restricted mobility, Townsend deprivation score)?
What is the incidence of recorded treatments, management and referrals made for incontinence in patients with and without dementia, stratified by sex and age?
Which covariates affect this relationship (polypharmacy, restricted mobility, Townsend deprivation score)?
Methods
Study design
A cohort study
Data source
Data will be taken from THIN (URL: http://www.epic-uk.org/thin.htm) covering general practices in the United Kingdom (UK) providing data during the period 1 January 2000 – 31 December 2009. This electronic recording scheme is one of the largest UK sources of continuous primary care data on patients’ consultations and prescribing data. It has been widely used for epidemiological studies, including a study of dementia and survival.13 Anonymised patient data are pre-collected from participating practices. Practices are broadly representative of UK general practices in terms of patients’ age and sex, practice size and geographical distribution. GPs enter medical diagnoses and symptoms using Read codes, a hierarchical recording system used to record clinical summary information. The age, sex, medical diagnosis and symptom records, health promotion activity, referrals to secondary care, prescriptions and quintiles of Townsend deprivation score are recorded for each registered individual.
Inclusion/exclusion criteria
In order to examine the incidence of incontinence (research question 1) we will first identify an ‘exposed cohort’ of individuals aged 60 or above with data available for at least 6 months between 1 January 2000 and 31 December 2009. Any patients with Read codes indicating learning disabilities and specifically Down’s syndrome will be excluded as this is a known risk factor for both dementia and incontinence. Based on the overall age and gender distribution in this cohort we will identify a comparison ‘unexposed cohort’ stratified by age, sex and practice, by randomly selecting from patients over the age of 60 without a record of dementia.
For all research questions, data will be taken from each practice after they have achieved an acceptable level of data quality as defined by the Acceptable Mortality Rate (AMR) date.409 We will use established methods to identify and exclude likely prevalent cases of incontinence or dementia recorded within 120 days of registration. This cut-off has been established by analysing THIN using the method of Lewis et al., specifically for incontinence and dementia.403 Pre-existing incontinence and/or dementia will be identified from any time point in the patient’s data, except for ‘incontinence’ codes prior to age 16.
Study variables
Dementia and incontinence will be identified from Read codes and drug codes, excluding single isolated drug codes and any codes flagged on THIN as having errors. Learning disabilities, Down’s syndrome and restricted mobility will similarly be identified from Read codes. Lists of Read codes have been developed in order to identify patients with these conditions. Codes were identified by searching the code dictionary for relevant terms (e.g. continen*, faecal, urin*, dement*, alzheimer*, memor*), then adjacent codes were examined and searches run on other keywords which are suggested by the dictionary definitions. This process has been used on a number of studies using THIN data.410 Drug/device codes were identified in the same way. All these lists were checked by two clinicians for relevance and completeness.
The first date against a code for dementia, incontinence and restricted mobility will be stored, allowing dementia and mobility to be modelled as time-varying predictors. The first date of each of various types of treatment/referral for incontinence will also be stored for use in research question 2.
We will use drugs prescribed as a proxy index for co-morbidity, calculated from the therapy records by counting the number of distinct BNF sections prescribed within the 6 months preceding the first incontinence code. For those without incontinence, a random date will be selected within their time at risk (uniformly distributed) and 6 months of prescriptions taken prior to that point. These calculations will apply to both exposed and unexposed cohorts.
Year of birth and sex will be extracted from the patient data files. Townsend deprivation scores (in quintiles) will be extracted from the postcode variable information (PVI) file for each practice.
Other variables to be extracted include:
practice and patient ID
indicator variables for the presence of dementia, incontinence (including pre-existing codes), and different types of treatments or referrals
the AMR date for the practice
year of birth, registration date, transfer date and death date.
Analysis
For research question 1 (incidence of incontinence in dementia compared to no dementia), incidence rates will be derived for incontinence in people with and without pre-existing dementia, and stratified by age, gender and deprivation. Epidemiological calculations of rate ratios will be used for the stratified analysis. Continuous covariates and inter-practice variation will be accounted for in a multilevel Poisson regression model. Incontinence will be the outcome, dementia the exposure, potential covariates are age, sex, deprivation, polypharmacy, and calendar period, and in a Poisson model time at risk will be an offset variable. This will allow us to find the adjusted relationships between dementia and incontinence and to consider whether the covariates interact (are effect modifiers) with dementia. We will account for variation between practices by modelling this as a random effect in the Poisson regression. Results will be compared with published statistics that are representative of current practice.116
For research question 2 (access to, and timeliness of, treatment for incontinence in dementia compared to no dementia), another multilevel regression model will be constructed with treatment options for incontinence as the outcomes, dementia as the exposure, and covariates defined as for question 1. A multilevel Poisson regression will be used and the offset and random effect will be defined as above. This will allow us to quantify any differences in access to treatment between those with and without dementia.
Limitations
The data are limited to consultations recorded in general practice and there may be differences in coding by different GPs, which we will investigate where possible. Because the data are drawn from GP consultations, there could be a Berkson’s bias inflating associations between morbidities. This will be minimised by looking over long time periods and measuring the time to first recording (incidence) of a diagnosis or treatment.411 We also know that people with continence problems can delay seeking help, sometimes by years, so there may be a bias toward earlier recording of incident incontinence in patients with dementia. We will cross-check diagnostic codes against prescriptions and free-text comments for under-recording or delayed recording of diagnoses.
