Chapter 7. The "Screening" for Disability Stage

7.1 Approaches to Screening

Some of these general advantages and disadvantages of Census versus survey approaches clearly carry over when we focus specifically on the “screening” stage of the exercise, where the aim is simply to identify a sub-set for more intensive investigation. The UN Statistical Division recommended in Principles and Recommendations for the 2000 Population and Housing Censuses the following list of broad categories of disabilities (based on the ICIDH) to be included when enquiring about disability:

  • seeing difficulties (even with glasses, if worn);
  • hearing difficulties (even with hearing aid , if used);
  • speaking difficulties (talking);
  • moving/mobility difficulties (walking, climbing stairs, standing);
  • body movement difficulties (reaching, crouching, kneeling);
  • gripping/holding difficulties (using fingers to grip or handle objects);
  • learning difficulties
  • behavioural difficulties
  • personal care difficulties
  • other (specify)

On the basis of UN recommendations, if a person indicates having one or more of the difficulties reported in the list, he or she is then identified as having a disability (1997: 93-94). The predecessor to the ICF provided the conceptual framework from which the list was drawn. By so doing, these recommendations made an attempt at standardisation and the use internationally recognised terminology. It is important to note that when the purpose of using the census is to establish a sampling frame for a more detailed survey, then these guidelines state that a generic question may be used. Mbogoni and Me (2000:8) point out, that although these recommendations were developed prior to the revision of the ICF, the recommended categories are still included. However the items on seeing, hearing and behaviour are now part of a different component of functioning - Body Functions rather than Activities.

Mbogoni and Me (2002) attribute international differences in disability prevalence rates found in censuses in part to methodological differences regarding: the definition of disability, the design of questions and the types of disabilities included. Examples are sited of countries whose definition focuses exclusively on one of the following; activity limitations; participation restrictions or impairments. It must not be forgotten that cultural practices and perceptions also play an important part. They distinguish between 3 types of screening question:

Question Type 1: A general question on the presence of the condition combined with items on participation and activity limitation, used in Canada, Australia, Caribbean region.

Question Type 2: a generic question on the presence of a disabled person in the household followed by a question asking what type of disability the person has( can be open or a scale).

Question Type 3: A checklist of impairments. This type has been included in census questionnaires in Africa, Asia and South America.

These authors highlight the lack of uniformity at the international level in question type resulting in varying prevalence rates: They conclude that the use of Type 1 questions brings with it the highest prevalence rates, while the lowest rates are reported where type three is in use.

7.2 Testing Census Screening Questions

The Australian Bureau of Statistics and Statistics Canada have carried out census tests on screen questions since the early 1990’s. The following criterion were used to test reliability of disability screen questions:

Census response negative - interview response positive = false negative

Census response positive - interview response negative = false positive

Two important criteria which would merit inclusion in the census were:

  1. high rate of matched positives (80%) on the census test and the follow up survey
  2. a close relationship between the census disability rate and the follow up survey disability rate. Census disability rate/ survey disability rate = 1 or close to 1.

Although much work was carried out in Australia (focus groups developed to test different versions of questions) conclusions from the last test in 1998 were that further work was needed in order to include a screen question in the 2006 census.

Statistics Canada used 2 filter questions in their 1986 and 1991 census surveys. Results showed the exclusion of a significant proportion of the population and Statistics Canada set about redesigning these questions.

The census screen used for the Participation and Activity Limitation Survey (PALS) in Canada, focused on activity limitations and participation restrictions related to health conditions (Langlois, 2002). The screen permitted an estimate of the number of persons experiencing limitations in their daily activities or restrictions in their participation in various settings, but provided little information at the level of body structures or body functions. Langlois notes that there are constraints on a global or screening measure even at the level of activity and participation: “it is difficult to cover all domains because of the necessary brevity of the question and some domains are simply not appropriate to include in some surveying contexts”. (2002, p. 5). The PALS census screen, for instance, did not refer to personal care or to interpersonal relations because of their sensitivity.

The filter questions differed from earlier forms used in the Canadian Census in that negative terms such as ‘limit’ were replaced by more neutral terms such as ‘reduce’. Prior to the census, a test was conducted (the 1999 HALS Test) with 12,500 respondents to check the extent to which this filter tended to exclude people who, on a more detailed set of questions, might be classified as ‘having a disability’. The result of the test indicated that the new filter represented an improvement, compared to the 1991 filter, in including a higher proportion of respondents who, on the more detailed set of items, would be classified as having a 'severe disability'. A pilot test conducted in 2000 found that only 2.4 per cent of respondents responding negatively to the screen had a ‘severe disability’, compared to 20 percent of those responding positively.

7.3 Design of Filter Stage for the Current Study

In order to pilot the main questionnaire for the current project we required a sample size of about 100 persons (composed of both adults and children) who had an activity limitation according to the definition applied. This required that we undertook a sifting of the national population to identify relevant respondents for the main survey. The most important design aspect at this sifting stage was choosing filter questions which would capture our target population.

Typology of Screen Questions – Q1 to Q2c

Questions 1 and 2 below were the filter questions used to target our sample (see Appendix 2). The questions chosen had their basis on the disability filter questions used in the 2001 Canadian Census which, as discussed above, was extensively tested. These questions are shortened versions of measurements of activity and participation restrictions for inclusion in a census. It was felt that these questions could be applied to a national sample. Although a generic set of questions, the broad ICF components are clearly apparent. As can be seen in Table 7, taken together the questions focus on activity limitation and participation restriction. The focus of Q1 is on activity limitation denoted by the domains of hearing, seeing, communicating, walking, climbing stairs, bending and learning. Question 2a to 2c focus on participation restriction with regard to the broad areas of home, work, school, leisure while also touching transport, which comes under environmental factors in the ICF. All of these questions include measures of frequency (yes sometimes, yes often, no) rather than a simple binary response (yes/no).

Q1. Does anyone in your household have any difficulty hearing, seeing, communicating, walking, climbing stairs, bending, learning or doing any similar activities? This person could be yourself, or a child or an adult of any age who is living in your household.
  • Yes sometimes
  • Yes often
  • No
Q2a. Does a physical condition or mental health condition or health problem reduce the amount or the kind of activity you or anyone in your household, can do…
  1. At home?
    • Yes sometimes
    • Yes, often
    • No
  2. At work or at school?
    • Yes sometimes
    • Yes, often
    • No
  3. c) In other areas, for example, transport or leisure?
    • Yes sometimes
    • Yes, often
    • No

The ICF was used as a conceptual framework in the design of these questions. The components, chapters, blocks and item level description are outlined in Table 7 below:

Table 7: Classification of the screen questionnaire according to ICF
ComponentSub-componentChapterBlockItem level

Body Functions

Hearing

2

Sensory functions and pain - hearing functions

b230

Body Functions

Seeing

2

Sensory functions and pain - seeing functions

b210

Activities & Participation

Communicating

3

Communication receiving-producing

d310 - d345

Activities & Participation

Walking

4

Mobility - walking and moving

d450

Activities & Participation

Climbing stairs

4

Mobility - walking and moving

d4551

Activities & Participation

Bending

5

Mobility - changing and maintaining body position - bending

d4105

Activities & Participation

Learning

1

Learning and applying knowledge - basic learning

d130-159

Activities & Participation

At home

6

Domestic life

d610 - d660

Activities & Participation

At work or school

8

Major life areas - education and work and employment

d810 - d839 and d840 - d859

Activities & Participation

Leisure

9

 

d920

Environmental Factors

Transport

5

Services, systems and policies

e510 - e599

Other Questions asked in the Filter questionnaire included:

Household Level

  • Total number of persons living in the household;
  • The number of persons aged over 18 living in the household
  • Whether there was more than one person in each household who had any of the difficulties or limitations referred to in the filter questions.

Where more than one person with a disability was found in any given household the so-called “next birthday rule” was applied for random selection purposes (i.e., the individual whose birthday is coming up next is selected). Once this person was identified the filter or sifting questionnaire was filled out in respect of that person.

Individual Category

  • Age group
  • The main condition causing the persons difficulty or activity limitation
  • Willingness to participate in the main survey
  • If not what was the main reason for not participating
  • Contact details for the main survey.

7.4 Piloting the Filter Questionnaire - July 2003

Methodology

In July 2003, the filter questionnaire was administered by telephone to a sample of 953 households by trained ESRI interviewers from the ESRI’s Offices in Dublin. Counties surveyed in the course of the pilot were Dublin; Westmeath, Meath; Mayo and Kerry. These were selected to ensure a reasonable rural/urban mix. Random digit dialling was used. A random sampling point was chosen in each of the five counties with 100 random phone numbers being generated in each cluster. The 100-bank of numbers in each sampling point was then used for the sifting process.

HOUSEHOLD LEVEL DESCRIPTIVES

Response Rates

As can be seen in Table 8 below a very high response rate was obtained in the filter survey. Out of the 951 households contacted by interviewers a total of 81 per cent (770) completed the survey with only 181 (19 per cent) refusing to do so. This response is surprisingly high for surveys of this sort. Response rate as high as this would more generally be associated with ‘face to face’administration. The questionnaire was, however, quite short taking on average 4 minutes to administer.

Table 8: Response Rates from Filter Telephone Survey – July 2003
 

N

%

Completed77081.0
Refused18119.0
Total951100.0

Screen Questions Q1 to Q2c

Household Prevalence Rate

Identification of the presence of a person with a disability in the 770 households which successfully participated in the pilot survey was based on the respondent answering in the affirmative to one or more of the four screen questions (Q1 to Q2c). This is taken to mean that one or more persons in the household had either an activity limitation, a participation restriction or a combination of both as a result of a health condition.

Table 9 presents details on the distribution of the 770 households which successfully completed the sifting questionnaire according to number of residents identified as experiencing a disability or related difficulty. From this one can see that 64.5 per cent of the sample (representing 497 households) were classified as having no-one with a disability/activity limitation on the basis of the definitions employed (i.e. they answered ‘no’to all four screen questions. A further 30 per cent of households (229 cases) recorded 1 person with a disability while the remaining 6 per cent of households recorded 2 or more persons.

Table 9: Distribution of households according to number of residents identified as experiencing a disability/difficulty

No of persons with disability

No of households

Per Cent of households

None49764.5%
1 person22929.8%
2 + persons445.7%
Total273100.0%

These figures mean that 83 per cent of all households in which someone has a disability contain only one member experiencing a disability/activity while those remaining 16 per cent of relevant households have 2 or more such individuals.

Individual Prevalence Rate

The total number of individuals resident in the 770 households which successfully completed the pilot sift was 2,439. A total of 317 of these was identified as having a disability. This gives a disability prevalence rate among individuals of the order of 13 per cent.

Disability prevalence levels with which to compare this figure can be extracted from the 2002 Census and also the module attached to the CSO’s Quarterly National Household Survey (QNHS) in the 2ndQuarter of 2002, as discussed earlier. The Census 2002 figure was 8.3% while the QNHS figure was 10%, but a number of important issues must be taken into account in comparing these three incidence rates (i.e. that derive from the Census of Population, the QNHS and the ESRI pilot exercise):

  1. differing survey methodologies two sample surveys and a census;
  2. differing sample sizes,
  3. The Census figure referred to the total population (institutional and non-institutional) while the QNHS and pilot figures related to private households only.
  4. the age group targeted in the QNHS was between 15 to 64 as opposed to all ages in the census and the NDA pilot survey;
  5. Significant differences could arise due to question typology and wording as discussed earlier. Questions used in the NDA pilot and the Census were based on measures of both activity limitation and participation restriction while the questions asked in the QNHS asked more specifically about impairment.
  6. The ESRI estimate is based on only a small and unweighted sample which is not intended to be representative of the population from which it was selected (i.e. of the 5 countries in which the pilot sift was undertaken).

Outcomes

Households in which a person with a disability was identified were asked whether or not they would be willing to participate in a personally administered follow up survey. There were three possible outcomes from the survey as outlined in Table 2.4. This shows that a total of 35 per cent of households which successfully completed the sift recorded that they contained a person with a disability. Of this 54 per cent (143) were willing to take part in the main survey and provided contact details while 46 per cent, (123) were unwilling to take part.

Table 10 outlines the reasons given by the 123 households identified as having someone who experienced a disability but where the household did not want to participate in the main survey (i.e. the 123 households in Table 2.4 which completed the sift; recorded that they had a resident who experienced a disability or related difficulty but which declined to provide contact details and to participate in the main survey).

Table 10: Outcome Categories figures
 

N

%

1. Completed/Disability no contact12316%
2. Completed/Disability with contact details14319%
3. Completed/No disability50465%
Total:770100%

One can see from the table that the main reason given was inability to participate (24 per cent), followed by the perception that taking part would be of no benefit (21 per cent). Between 14 to 16 per cent of respondents unwilling to take part cited reasons of lack of time/interest and shyness/ not wanting to take part. Where other reason was given no reason was given and a small number were too ill or in hospital.

Table 11: Reasons for not participating
 

Number

%

Lack of time2016%
Lack of interest1714%
No benefit2621%
Unable to participate2924%
Shy/not wanting to1915%
Other reason1210%
Total123100%

Characteristics of individuals identified as experiencing a disability/activity limitation

Age

We now turn to a consideration of characteristics of the individuals identified in the course of the sift as having a disability/activity limitation in terms of age and main health condition. Table 12 provides a breakdown by age in respect of those willing to participate in the main survey as well as those who did not wish to participate. Although there would not appear to be any simple systematic relationship between age and propensity to participate in the main survey one can see that there is a 10 percentage point differential in the oldest age cohort in the proportions willing to participate in the main survey. [1]

Table 12: Age breakdown…
Age% Willing% Unwilling

Under 5

1%

2%

5 - 15

10%

6%

16 - 24

3%

8%

25 - 44

16%

12%

45 - 64

33%

26%

65+

37%

47%

Willing: N=143 Unwilling: N=123

Main Health Condition

Question 9 of the filter questionnaire recorded details on the main condition which causes a person’s difficulty or limits his/her activities. The health conditions recorded were classified using the International Classification of Diseases ICD-9. Table 13 presents a breakdown by health condition and whether or not they were willing/unwilling to participate in the main survey.

In interpreting the figures the reader is reminded of the small sample numbers involved. Notwithstanding this caveat, it would appear that there is really no major systematic difference in the breakdown of health conditions between those identified as willing/unwilling to participate in the main survey.

In general prevalence terms, one can see that the most frequently occurring disabilities/activity limitations are musculoskeletal problems (over 30 per cent). These relate to a variety of factors including age related issues such as arthritis, stiffening of the joints etc. The next most frequently occurring problem/difficulty related to the nervous system (20 per cent); mental disorders (15 per cent) and the circulatory system (10 per cent).

Table 13: Breakdown by main health condition of those willing/unwilling to take part
ICD-9 Broad Categories

Willingness to participate

Willing

Unwilling

% of Total

Infectious and Parasitic Diseases0.00.70.4
Neoplasms0.82.21.5
Endocrine1.74.43.0
Diseases of the blood1.71.51.5
Mental Disorders and Learning Disabilities15.815.315.4
Nervous System21.718.220.2
Circulatory System10.010.910.5
Respiratory System5.83.64.5
Genitourinary System1.51.50.7
Skin0.80.00.4
Musculoskeletal34.228.531.1
Congenital Anomalies1.72.92.6
Perinatal0.00.70.4
Symptoms/signs/ill defined conditions3.34.43.7
Injury & Poisoning0.85.13.0
Total100.0100.0100.0
(number)(143)(123)(251)
  1. This, of course, illustrates the importance of statistical reweighting or ex post re-adjusting of the effective sample prior to analysis. As noted above, given the scale and objectives of the pilot exercise no such re-adjustment was undertaken.