Epidemiology is the study of disease and disability in populations. Population is used in the technical sense as a group of subjects who all share a common characteristic. This may be that they all live in a certain location, the usual sense of the word, but other characteristics can be used to define a population, such as age, gender or occupation. In health related studies the health setting is often a characteristic, for example people attending a general practice or hospital.
Populations can be studied by a census or a survey. A census is the complete enumeration, and measurement of all subjects who constitute of a population. A survey is the measurement of only some of the subjects within a population. In both situations there is usually a target population, the population of interest. There is a ‘frame’, the technical name for the list of subjects in the target population, and finally the group of subjects who are actually contacted and measured. For various reasons the frame may not be efficient, in the sense that it doesn’t contain all the elements of the target population. For example if an electoral role is used as the frame for subjects living in an electorate, a particular geographical area, then some people may not have enrolled for the election. Also even using the frame there may be subjects who it is not possible to contact or, if contacted, it is not possible to measure them.
Non-response bias is the name given to the effect of frame inefficiency, inability to contact subjects and inability to obtain a useful response. It is a very important consideration in interpreting studies of disease. This is because studies are often concerned with making predictions concerning the disease in the whole population of interest based on information from a proportion of the population, that is most studies of disease are usually surveys. If the subjects who do not respond to the survey are different to those who do, this means the predictions made from the responses that were obtained will not reflect the ‘true’ state of affairs. An example from the political arena was a telephone survey of voter preference in the lead up to an American election in the 1940’s. There was a huge number of responses, over 1 million, predicting a victory for the Republican candidate. In fact the Democratic candidate won. The reason was that at that time only wealthy people had telephones and they were more likely to vote Republican. Here the target population was all voters. The frame was the telephone list and the two did not coincide.
While intuitively it seems that a census would be preferred under most circumstances, to reduce non-response bias by contacting all elements of a population, in fact surveys concentrating on samples of a population are often better. The reason for this is that the huge resources devoted to contacting all elements of a population are often better used in obtaining higher quality information from a smaller set of the population. This might be particularly so for a disorder such as urinary incontinence where people may give better information to an interviewer in person than to a mailed questionnaire. The important thing about surveys is that the sampling procedure, the way subjects are chosen for interview for example, must be free of another form of bias, selection bias. This is the term that describes how subjects may differ in a systematic way from the population of interest because of the way they were chosen. The best sample surveys use random selection procedures, usually based on a sample frame.
Some studies, including some of those discussed in the material later in this section, use a hybrid technique. That is they attempt to contact all elements of a group, a census, living in a particular geographic area and then extrapolate from this group to the national population.
Measurement of the subjects is also an important issue in the study of disease and disability. Measurement is carried out by means of ‘instruments’. For example a ruler is used to measure height. Here the characteristic of interest is the height and the instrument is the ruler. For disease and disability the instruments are often questionnaires. These can be self-administered, that is the subject fills them out without intervention of the investigators, and postal questionnaires are typically of this sort. Interviewers can administer instruments either by telephone or in person.
Problems with instruments include that they may not measure what they purport to measure. For example an instrument to measure urinary incontinence may not get a sensible response because the words ‘urinary incontinence’ may be meaningless to the subject. Another way this can happen is when measuring frequency of events. Often subjects have recall bias, so for example they may say they have wetting episodes five times a day when in fact this only happens once a week and they usually have incontinence once a day. In addition for urinary incontinence symptoms are subjective and may not correspond particularly well to underlying bladder and voiding disorders. Instruments may be subject to bias. This may occur at a subject level where the subject understands the question but gives a response that is socially acceptable or that reflects what the subject thinks the interviewer wants to hear. It may also occur at interviewer level where the interviewer asks questions in a way that encourages a certain answer. Different interviewers may get different responses. A questionnaire may not cover areas of interest and may leave little role for open-ended responses. In general questionnaires about urinary incontinence that are administered by an interviewer obtain higher prevalence rates than postal questionnaires.
For all these reasons then, different survey and census techniques may give different rates and types of urinary incontinence. An investigation into the reasons and consequences of non-response for a postal survey about urinary incontinence suggested older ‘non-responders’ may have a greater prevalence of ill health and urinary incontinence and so prevalence figures may under estimate incontinence in people aged 70 years or older (1).
Some more technical terms that are used in epidemiological studies include incidence and prevalence. Incidence is the number of new cases of a particular disorder over a period of time. A typical way this is expressed is number of new cases per 1000 population per year. It is important to think about which population the incidence is expressed for as incidence may vary with age, gender, parity (number of children), ethnic group and so on. How a case is defined may also be important. Prevalence is the total number of cases of a particular disorder present in a population at one particular time. Again it is important to specify which population and how a case is defined. Prevalence depends on not only incidence but also survival. A condition with a very low incidence but a long survival with the condition not improving may end up with quite a high prevalence. A condition with a high incidence but a very high rate of cure (or rapid death) may have a low prevalence.
Two types of study that are sometimes done to see how rates may vary with various factors include cross sectional studies and longitudinal (or cohort studies). In cross sectional studies a number of different populations are surveyed at the same time using the same instrument. If the prevalence rate varies according to the different characteristics of the sampled populations then this may mean that the particular characteristic might be causing the variation in prevalence. It may also mean however that there are ‘third’ variables that explain the difference and have nothing to do with the variable that was examined. Longitudinal studies follow a particular group of people over a period of time making measurements on them.
Urinary Incontinence: Epidemiology
Different surveys suggest different prevalence rates for urinary incontinence. Some of the reasons for variation in reported prevalence include the different populations surveyed, differing definitions of urinary incontinence, and different ways of surveying populations. For example surveys that use face-to-face techniques typically find higher prevalence rates than postal questionnaires.
A review that attempts to come up with a summary figure for the prevalence of urinary incontinence is summarised in the following table (2).
Three New Zealand studies were cited in the review by Thom (2).
The first study conducted by Campbell and colleagues was published in 1985 (3). This study looked at a random sample of people over the age of 65 from an electoral role in Gisborne. The definition of urinary incontinence for this study was whether ‘you ever wet yourself’. The study was carried out by face-to-face interview. 11.6% of the population gave a history of urinary incontinence. Of this group 64% had urinary incontinence either more than 2 to 3 times a week or had an indwelling urinary catheter. Figures were not given for those resident in institutional care except to note residence in care made urinary incontinence more likely. Urinary incontinence was also far more prevalent amongst those with dementia affecting about 50% of those with dementia, who in turn made up 15% of the sample.
The second study conducted by Holst and Wilson was published in 1988 (4). This study was a phoned interview of a sample of women from the electoral register living in Dunedin. Any incontinence and ‘regular’ incontinence, defined as urinary incontinence more often than twice a month were specifically asked for. The overall prevalence of urinary incontinence was 31%. Seventeen percent of women had ‘regular’ incontinence. In this study apart from the age group 18-24, who had a very low prevalence of urinary incontinence, the prevalence or regular incontinence was not different amongst the different age groups. Of interest in this study only 35% of women with regular incontinence had sought health professional advice.
The third study conducted by Lara and Nacey was published in 1994 (5). This was a postal questionnaire with phoned follow up of a sample of women from the electoral register living in some cental suburbs of Wellington. Any incontinence was used as a definition but the frequency of urinary incontinence was also recorded. The overall prevalence of urinary incontinence was 34% with a slightly higher prevalence in Māori, 47%, and European, 31%, than Pacific Island women, 29%. Overall 37% of the women with urinary incontinence had incontinence once or more daily, or 13% of the overall figure. It is interesting to note that these figures are very similar to those in the Dunedin study.
Two recent New Zealand studies give prevalence data by age for children for enuresis. The study by Ferguson (6) found that the prevalence of failure to gain bladder control was lower than that of nocturnal enuresis at all ages and that by age eight 3.3% of children had failed to gain bladder control and 7.4% still had nocturnal enuresis. The study by Feehan (7) showed that by age 11 only 6% of children still had nocturnal enuresis. Estimates based on the 2006 census there were about 612,000 children aged between 8 and 17 in New Zealand.
The table below is adapted from the Ferguson study (6).
Excellent prevalence data for urinary incontinence in institutions providing long-term care is available from a census of all Auckland Rest homes published as a report in 1995 (8). The data regarding urinary incontinence is summarised in the table.
Thus overall in rest homes 13% of people had once daily, or more, urinary incontinence, while the figure for private hospitals providing long-term care is 64%.
No New Zealand data are available to determine the frequency with which urinary incontinence is a reason for entering residential care. In an Australian study examining reasons for referral to an Aged Care Assessment Team 32% of 324 referrals had incontinence as one of their problems (9). The prevalence of continence problems in people aged over 65 years referred to an Assessment and Service Co-ordination service is 31% (10).
About 6% of the population aged over 65 lived in residential care, 30,000 people (11). The survey results published in ‘Disability Counts’ (12) also gives insight into the number of people with severe disability who live in the community. From estimates based on this survey 17,400 people, aged over 15, were rated as having severe disability and living in residential care, and 53,500 people aged over 15 were rated as having severe disability and living in house holds. 25,700 people aged over 65 and living in the community are rated as having severe disability. The vast majority of people who live in residential care are over the age of 65. If the pattern of disability is similar for those with severe disability who live in the community then it is reasonable to expect the prevalence of urinary incontinence to be similar also. From this study of the 27,600 people over aged 15 who live in residential care facilities, 87% are aged over age 65. In addition 10,300 of the 27,600, or 37%, use incontinence products.
The table shows the implications of the prevalence data for the number of people with urinary incontinence in NZ, based on the 2006 Census.
1Based on estimate from Ferguson (7) for children aged over age 8
2Based on population estimates for 2006, data from 2006 census by individual age not available
3Based on ‘ever incontinent’ estimates from Thom (1) for adults
4Based on actual population counts from 2006 census (available in 5 year age bands)
5Based on Auckland residential care study (9)
Numerous recent studies have explored the prevalence of incontinence across many countries. An example of a recently published study of this sort is by Irwin and colleagues, the ‘EPIC’ study (13). The subjects for this study were recruited by a random sample of households with a residential telephone number. Questionnaires based on the 2002 International Continence Society definitions of continence and lower urinary tract symptoms were developed in the languages of each country and administered by phone. The five countries that participated were Canada, Germany, Italy, Sweden, and the United Kingdom. The summary figures for any urinary incontinence are shown by age band in the following table.
The researchers for this particular paper commented that the prevalence rates were at the lower end of the prevalence rates found in other studies and speculated that this may have been that definitions of urinary incontinence, questionnaire design, study populations, and survey methods vary across studies.
Besides trying to determine prevalence non-experimental observational studies are often concerned with identifying features of participants with incontinence that distinguish them from participants without incontinence. These differences could then be reasonably thought to be causative for the incontinence and potentially amenable to interventions. Trying to make statements about causation from non-experimental studies has many potential problems. These include confounding, a technical name meaning that although a feature is associated with incontinence the actual cause is something that the feature that hasn’t been measured. Another problem is ‘reverse causation’. For example people with stress incontinence may have a higher ‘risk’ of having had surgery for pelvic organ prolapse, so one could say that the surgery causes the prolapse. In fact the surgery was done in part because of the incontinence.
There are many studies looking at so-called risk factors that are potentially explanatory factors for incontinence. A set of these studies, for women, was based on the Norwegian ‘EPINCONT’ study (14-). The paper describing prevalence, symptom type, and the impact of the incontinence found prevalence increased from around 16.6% in the age group less than 40 years of age, to around 27.5% of women aged 40 to 59 years, and 29.9% aged over 60(14). One third of women described their incontinence as causing some bother (24%), much bother (6%), and a great deal of problems (4%). The number of children seemed to affect continence rates, increasing with the number of children, below age 65 but not in older women (15). Having a child below the age of 25 was associated with a lower risk of incontinence (16). Heavy smoking, obesity, and tea drinking but not high intensity physical activity, coffee or alcohol consumption were associated with increased incontinence (17). Caesarean section seems to protect against incontinence compared to vaginal delivery (18). Risk factors for male incontinence are not well established.
Although older age is itself a risk factor for urinary incontinence even amongst older people it is useful to know what factors are associated with urinary incontinence and so may be causative, and hopefully reversible. Factors associated with urinary incontinence in this group were older age, particularly age over 85 years, worse activity of daily living, cognitive function, the presence of diabetes, a history of stroke, the presence of delirium, environmental barriers (including physical restraints) and a urinary tract infection (although how this was diagnosed was not clear) (19). For people who live in residential care factors associated with urinary incontinence include problems with activities of daily living, dementia, use of restraints (including bed rails) and the use of anti-anxiety/hypnotic medications (20).
Thom D. Variation in estimates of urinary incontinence prevalence in the community: Effects of differences in definition, population characteristics, and study type. J Am Geriatr Soc 1998;46:473-480
1. Dallosso HM, Matthews RJ, McGrother CW, Clarke M, Perry C, Jagger C. An investigation into nonresponse bias in a postal survey on urinary symptoms. BJU International 2003;91:631-6
2. Thom D. Variation in estimates of urinary incontinence prevalence in the community: Effects of differences in definition, population characteristics, and study type. J Am Geriatr Soc 1998;46:473-480
3. Campbell AJ, Reinken J, McCosh L. Incontinence in the elderly: Prevalence and prognosis. Age Ageing 1985;14:65-70
4. Holst K, Wilson PD. The prevalence of female urinary incontinence and reasons for not seeking treatment. NZ Med J 1988;101:756-758
5. Lara C, Nacey J. Ethnic differences between Māori, Pacific Island and European New Zealand women in prevalence and attitudes to urinary incontinence. NZ Med J 1994;107:374-376
6. Fergusson DM, Horwood LJ, Shannon FT. Factors related to attainment of nocturnal bladder control: An 8 year longitudinal study. Pediatrics 1986;78:884-890
7. Feehan M, McGee R, Stanton W, Silva PA. A 6 year follow up of childhood enuresis: Prevalence in adolescence and consequences for mental health. J Paeditr. Child Health 1990;26:75-79
8. Broad JB, Richmond DE, Bonita R, Baskett JJ. Changes in long term care of older people in Auckland between 1988 and 1993. Auckland: Academic Section of Geriatric Medicine, University of Auckland, 1995
9. Cape RDT, Gibson SJ. The influence of clinical problems, age and social support on outcomes for elderly persons referred to regional aged care assessment teams. Aust NZ J Med 1994;24:378-85
10. Weatherall M, Slow T, Wiltshire K. Risk factors for entry into residential care after a support needs assessment. NZ Med J 2004;117:U1075
11. Statistics New Zealand website http://www.stats.govt.nz/ (Accessed 31 January, 2007)
12. Disability Counts. Statistics New Zealand. Statistics New Zealand, Wellington, 1998
13. Irwin DE et al. Population-based survey of urinary incontinence, overactive bladder, and other lower urinary tract symptoms in five countries: Results of the EPIC study. European Urology 2006;50:1306-15
14. Hannestad YS et al. A community based epidemiologic survey of female urinary incontinence: The Norwegian EPICONT study. J Clin Epidemiol 2000;53:1150-7
15. Rortveit F et al. Age- and type-dependent effects of parity on urinary incontinence: The Norwegian EPINCONT study. Obstet Gynecol 2001;98:1004-10
16. Rortveit G, Hunskaar S. Urinary incontinence and age at the first and last delivery. Am J Obstet Gynecol 2006;195:433-8
17. Haanestad YS et al. Are smoking and other lifestyle factors associated with female urinary incontinence? The Norwegian EPINCONT study. BJOG 2003;110:247-54
18. Rortveit G et al. Urinary incontinence after vaginal delivery or Cesarean section. N Engl J Med 2003;348:900-7
19. Landi F et al. Potentially reversible risk factors and urinary incontinence in frail older people living in the community. Age Ageing 2003;32:194-199
20. Brandeis GH et al. The prevalence of potentially remediable urinary incontinence in frail older people: A study using the minimum data set. J Am Geriatr Soc 1997;45:179-184
