Aging Clin. Exp. Res. 5: 371-383, 1993
The measurement of Instrumental ADL: Content validity and construct validity K. Avlund 1 ,2, K. Schultz-Larsen 1 ,2, and S. Kreiner3 lInstitute of Social Medicine, University of Copenhagen, Copenhagen, 2The Glostrup Population Studies, Glostrup University Hospital, Glostrup, 3Danish Institute of Educational Research, Copenhagen, Denmark ABSTRACT. A new measure of Instrumental Activities of Daily Living (IADL), which is able to discriminate among the large group of elderly who do not depend on help, was tested for content validity and construct validity. Most assessments of functional ability include Physical ADL (PADL) and Instrumental ADL (IADL). PADL-scales assess the basic capacity of persons to care for themselves. IADL-scales are used to assess somewhat higher levels of performance, such as the ability to perform household chores or go shopping. Data were collected from 734 70-year-old people in Denmark in the county of Copenhagen. The measure of Instrumental ADL included 30 activities in relation to tiredness and reduced speed. Construct validity was tested by the Rasch model for item analysis; internal validity was specifically addressed by assessing the homogeneity of items under different conditions. The Rasch item analysis of IADL showed that 14 items could be combined into two qualitatively different additive scales. The IADL-measure complies with demands for content validity, distinguishes between what the elderly actually do, and what they are capable of doing, and is a good discriminator among the group of elderly persons who do not depend on help. It is also possible to add the items in a valid way. However, to obtain valid IADL-scales, we omitted items that were highly relevant to especially elderly women, such as house-work items. We
conclude that the criteria employed for this IADL-measure are somewhat contradictory. (Aging Clin. Exp. Res. 5: 371-383, 1993)
INTRODUCTION Activities of daily living The assessment of functional ability is based primarily on the ability of individuals to manage activities of daily living (ADL) at home, at work and in their leisure-time (1,2). Most of these assessments include Physical ADL (PADL) concerning basic bodily maintenance (3), and Instrumental ADL (IADL) concerning activities fundamental to independent community residence (4-7). Most studies which discuss ADL are concerned with disabled elderly who are in need of help (8, 9). The purpose of the present study was to develop a generally sensitive measure of functional ability among the large group of elderly who do not depend on help, and thus obtain a suitable functional ability measure for health studies of community dwelling elderly, in particular as a summary statement of the individual's health status. Three important criteria were employed for this measure of IADL: 1) It should include a broad range of activities that are relevant to elderly people, and the ability to perform these activities should be measured by classification categories relevant to elderly
Key words: Construct validity, content validity, functional ability, Instrumental Activities of Daily Living, Rasch model. Correspondence: Kirsten Avlund, Institute of Social Medicine, University of Copenhagen, Panum Institute, Blegdamsvej 3, 2200 Copenhagen N, Denmark. Received April 22, 1992; accepted February 2, 1993.
Aging Clin. Exp. Res., Vol. 5, No.5 371
K. Au/und, K. Schultz-Larsen, and S. Kreiner
people. Furthermore, it should be able to measure activities of daily living among both men and women in the same way, and it should distinguish between what the elderly actually do and what they are capable of doing. This way the measure should comply with demands for content validity (10, 11). 2) It should be able to discriminate among the entire group of elderly people who do not depend on help (12-14), and it should be able to distinguish between different groups of elderly with regard to other health indicators. With the intention of achieving a greater variability and a larger capacity for discriminating among a normal population of elderly persons, we included a higher than usual number of activities related to ADL and used two newly developed classification levels. 3) It should be possible to add the items in the IADL-measure to an aggregate index in a valid manner (15-17). This is associated with two main problems. The first problem is qualitative: do the items belong to the same dimension? The second problem is quantitative: how should response categories be scored, and items weighted in the index scales? (18-21). None of the existing IADL measures have met all these three criteria. A previous study based on the Glostrup Survey in 1984 (22) addressed what 70-year-olds in a random sample actually do and what they are capable of doing. Major differences were found between PADL and IADL. In considering what the elderly actually do, it was found that: 1) all PADL were performed by nearly everyone, whereas IADL were not performed by everyone; and 2) there were significant differences between sexes with regard to nearly all IADLitems. The latter was not the case for the PADL. In considering what the elderly are capable of doing, it was found that only a few were incapable of performing the PADL, while a larger number were incapable of performing the IADL without help. These differences between PADL and IADL may be due to the fact that not all IADL are relevant to everyone, as one's IADL depend on gender, cultural setting, housing-facilities, economic situation, and interests. This was convincingly demonstrated in surveys of functional abili-
372 Aging Clin. Exp. Res., Vol. 5, No.5
ty among elderly people from eleven nations (23-25). The results showed large differences in ability to manage ADL, especially IADL. Substantial differences between different areas were found; for instance, 90% of the 80-84-year-old men in Kuwait had difficulty performing even light housework, us only 20% of the 80-84-yearold men in West Berlin. This strongly suggests that instrumental items are culturally influenced. The purpose of this study is three-fold: describe the ability of a random sample of 70year-old men and women to perform IADL; test whether the new measure of IADL forms a unidimensional structure, and assess the content validity of the measure.
MATERIALS AND METHODS This investigation is part of the 1984 study of 70-yearolds in Glostrup, Denmark (22, 26, 27). The study population described here comprised 804 men and women born in 1914 who participated in a comprehensive medical survey at the Copenhagen County Hospital in Glostrup in the period from April 1, 1984 to March 1, 1985. One to two weeks after the medical examination, an occupational therapist visited 734 of the participants in their homes (participation rate 91%), and on that occasion all 734 were interviewed about their functional ability and social situation. Non-participants did not differ essentially from those who participated regarding social and demographic characteristics, or days spent in hospital (27). The analyses described in this study include the following variables: - Social Economic Status (SES) was defined by latest occupation (22). - Chronic diseases were defined by international standardized patient reports of characteristic types of symptoms (e.g., intermittent claudication, bronchitis, angina pectoris). Patients were designated as having: a) intermittent claudication if they had pain in the lower extremities when climbing stairs and walking straight on; b) chronic bronchitis if they had both cough and sputum production on most days for at least 3 months during the previous 2 years; c) angina
The measurement of Instrumental ADL
pectoris if they had chest pain when climbing stairs, and walking straight on. The diagnosis of osteoarthrosis was based on a medical evaluation by the project physician that included recording of joint impairments in the lower limbs. - General symptoms. People were designated as having: a) dyspnea if they were breathless when climbing stairs, walking straight on at normal speed, walking straight on at their own speed, and when bathing and dressing; b) dizziness if they suffered from dizziness a couple of times per week or constantly. Walking-impairment was assessed by the question: "Are you impaired when walking?" General tiredness was assessed by the question: "Do you feel tired at present?" - Utilization of health and social services Hospitalizations were evaluated asking: "Have you been hospitalized during the last 12 months?" Homehelp was assessed by asking whether the participant had home help or not at the moment. Technical aids were evaluated by asking whether the participant used technical aids when performing daily activities.
- IADL: To satisfy the criterion of discriminating among all elderly, we included a higher than normal number of activities related to ADL, and three classification categories that are especially related to elderly people. The measure of Instrumental Activities of Daily Living thus comprised 30 activities related to transportation, writing, using the telephone, cooking, housework, doing the laundry, shopping, managing economy, gardening, snow shovelling, and needlework. The activities that the participants actually perform were described according to three different dimensions: 1) Whether the person's speed was as usual or reduced, compared to earlier in life. 2) Whether the person was or was not tired afterwards. 3) Whether the person did or did not need help. If a participant did not perform an activity, because it was irrelevant, it was coded as "not an-
swered". This means that all answered variables indicate what the elderly actually did or would do if they were able to, while the not answered variables indicate what the participants did not do for different reasons. As it was confirmed that the percentage of men and women who are not able to manage the individual IADL tasks without help was small, only reduced speed and tiredness were considered as part of the IADL scales. Thus, the cases where respondents were not able to manage without help were supposed to reflect both reduced speed and tiredness afterwards. We obviously loose information about the very disabled in this way, but this was accepted because the purpose of developing the method was to describe function among the huge group of elderly who are not dependent on help. Thus, the answers to the individual items constitute a raw score that summarizes responses to two dichotomous items (+slower speed and +tiredness), and the scales are calculated as the direct raw score. Consequently, the measurement of IADL does not include 30, but 60 different items. The plan of the analysis a) Analysis of the F-scale Since Fillenbaum's IADL screener (4) is known to be short, valid and reliable, we started the analysis by including the almost similar items in our preliminary scale analysis (the F scale). Figure 1 - Items in Fillenbaum's IADL-screener and in the F-scale.
Items
Fillenbaum's lAD L-screener
The F-scale
Transportation
Bicycle Public trasportation Prepare cold meals Vacuum clean Manage economy Shop
Cook Housework Manage economy Shop Categories
With or without help
With or without reduced speed With or without tiredness
Aging Clin. Exp. Res., Vol. 5, No.5 373
K. Avlund, K. Schultz-Larsen, and S. Kreiner
A traditiona1 Rasch item analysis of the items in the F-scale was carried out. The Rasch model is a latent trait analysis developed for construct validation that specifically addresses the internal validity by assessing the homogeneity of items under different conditions (28-30). It postulates that the response patterns, defined by answers to different questions or items, are homogeneous across items and individuals, and in this way it investigates whether the separate questions belong to the same dimension. The Rasch item analysis basically makes the same
demands on a scale as does the Guttman Scalogram Technique, that is, the demand of homogeneity between the included items and the included persons (31). In a perfect Guttman scale, however, the answering pattern is completely determined whereas the Rasch item analysis allows random variation. b) Inclusion of extra items To obtain a broader description of the entire range of function, we extended the F-scale by adding more activities to the scale, one by one.
Table 1 - Percentage of men and women who get tired, have slower speed and need help when performing the Instrumental Activities of Daily Living. (n) indicates the number of persons for whom the activity is relevant.
IADL items
Men N
Women N
Go by bicycle Use public transportation Prepare cold meals Vacuum clean Manage economy Shop Write letters Write memos Write own name Use the telephone Answer the telephone Cook for guest Prepare hot meals Make coffee/thea Cut bread/meat/vegetables Open cans, etc. Light housework Wring out a dishcloth Clean the floors Wash clothes by hand Use the washing machine Go to the launderette Paint Knock nails into the wall Garden work Shovel snow Darn the stockings Sew on the sewing machine Needlework Knit
248 231 171 202 310 279 185 323 361 290 319 67 117 222 210 207 131 132 113 82 96 79 265 290 271 215 41 20 55 1
231 331 366 350 311 349 257 356 359 356 360 339 358 368 368 368 365 365 352 350 328 176 135 94 208 93 328 270 345 296
374 Aging Clin. Exp. Res., Vol. 5, No.5
Tiredness Men Women % % 13 2 4 9 10 18 1 2 2 0 0 3 3 2 3 3 2 2 11 1 2 11 18 14 30 27 2 5 2 0
18 3 8 25 16 34 0 2 1 0 0 40 9 3 7 11 9 8 33 6 7 21 22 7 47 43 8 7 9 13
Slower speed Men Women % %
Need help Men Women % %
34 3 9 24 14 24 1 4 4 1 1 11 9 7 7 8 13 7 20 7 8 16 34 30 46 40 2 5 2 0
16 4 1 4 2 3 4 2 1 1 1 3 2 1 1 1 4 3 7 1 4 5 10 8 5 9 5 10 3 0
34 7 14 40 24 41 5 3 3 1 1 20 13 8 14 19 42 21 48 15 14 30 29 15 59 53 9 8 11 13
23 3 0 13 4 6 1 1 1 1 1 5 2 1 3 8 3 3 12 2 4 9 17 13
6 18 6 6 5 13
The measurement of Instrumental ADL
The purpose of these analyses was twofold: ensure that the extra item belonged to the same dimension as the items in the F scale, and estimate the item parameter of the extra item. c) Incomplete response patterns
The interviews about IADL distinguished between what elderly persons are able to do, and what they actually do. The consequence of this, however, is that we only have information about activities that are relevant to the individual person, and thereby have incomplete response patterns. The total study population comprised 366 men and 368 women. Table 1 reports the number of persons for whom the Instrumental ADL were relevant, and thus indicates the amount of incomplete answers. To minimize these problems, the Rasch analysis of IADL required special calculations. If only 6 of the 14 items were known, calculations were made to describe how the last 8 items would have been answered had they been relevant to this individual person. These calculations were based on the knowledge of how the 6 answered items were managed; if these 6 activities were managed optimally, our hypothesis based on the Rasch analysis is that the remaining 8 items would also be managed optimally, were they relevant for that person. If a person had very few answered items, the result of such a calculation will obviously be doubtful. These weaknesses are preferred, however, to an IADL-measure including activities that are not relevant to an individual person. It should be emphasized, however, that in order to obtain a good scale, most persons should have answered most of the items. The last step was to allow a scale-calculation in relation to incomplete response patterns: incomplete response patterns were completed by the technique suggested by Cook and Eignor (32), where the available information about both personal characteristics and response patterns on other items was used to complete the incomplete response patterns. An overview of some of the statistical techniques is presented in Appendix 1. d) Correlations between the two IADL scales and other health measures were analyzed by Pearson's X2 tests.
RESULTS
Main results Table 1 shows how the 70-yearolds manage the included IADL-activities. The percentage of men and women who are not able to manage the individual IADL-tasks is small. 74% of the women and 78% of the men manage all these activities without any need for help. Many 70-year-old men and women get tired, and have reduced speed when performing daily activities. 24% of the men and 41% of the women feel both tired and slow when they perform at least one of these IADL-tasks. The Rasch analysis of IADL showed that the proposed IADL measure forms two additive IADL-scales, of which the first is a tirednessscale of 22 items, and the second a reduced speed-scale of 17 items. a) Analysis of the F-scale With regard to F scale items, two scales emerged from the analysis: one in relation to reduced speed (bicycle, use of public transportation, preparation of cold meals, vacuum cleaning, managing economy), and one in relation to tiredness (bicycle, use of public transportation, preparing cold meals, managing economy).
b) Inclusion of extra items Five items did not fit into any of the scales, and were therefore excluded (shopping, preparing hot meals, washing the floor, knitting, needlework). Table 2 shows the two scales: a reduced speedscale, including 17 items and a tiredness-scale including 22 items. In the final analyses of this study, we chose to include only the items that belong to both the tiredness- and the reduced speed-scale in the IADL measure. The reasons for this procedure were 1) it is the most manageable, 2) the scale consisted of many items in any case, and 3) it was thus possible to compare the two categories of tiredness and reduced speed. Therefore, the final IADL-scales considered in this study include the 14 activities shown in Table 3. c) Incomplete response patterns
The results of the scale calculations with regard to the incomplete response patterns are seen in
Aging Clin. Exp. Res., Vol. 5, No.5 375
K. Aulund, K. Schultz-Larsen, and S. Kreiner
Table 2 - The items in the two IADL-scales. The marked "x" indicates that the item was accepted by the Rasch item analysis. Reduced speed Tiredness
F-scale Bicycle Use public transportation Prepare cold meals Vacuum clean Manage economy Extra items Use the telephone Receive telephonecalls Write letters Write memos Write own name Cook for guests Make coffee/thea Cut bread/meat/vegetables Open cans Light housework Wring out a dishcloth Wash clothes by hand Use the washing machine Go to the launderette Paint Knock nails into the wall Darn stockings Garden work Shovel snow Sew on the sewing machine
x x x x x x x
x x x x
x
x x x x x x x x x x x x x x x x x x
17 items
22 items
x x x x x x x x x
Table 3, which describes the mean scores of single items and the adjusted true scores which equal the final IADL-scale for complete response patterns. The observed mean score indicates the degree of ability of performing separate activities among those who actually do perform the activities in question, while the true IADL-score indicates the calculated degree of ability for the single items, as it would appear if the activities in question were relevant to everyone. A high mean score indicates a good functional ability, whereas a lower score indicates a poorer function. The order from the easiest to the most difficult is nearly the same in the raw score as in the final IADL-scale.
d) IADL in relation to other health measures Table 4 shows that there are strong correlations between functional ability on the two IADL scales and the other indicators of health. Gender differences Table 5 shows the distribution of men and women on the two IADL-scales. The persons with maximum score manage all 14 activities without reduced speed and without tiredness. Men and women are different with regard to IADL, as more men than women manage the 14 activities without reduced speed and without getting tired.
Table 3 - Scale calculation in relation to the incomplete response patterns in IADL. Reduced speed
Make coffe Use public transportation Prepare cold meals Wash clothes by hand Cut bread/meat/vegetables Sew on the sewing machine Use the washing machine Open cans/etc. Manage economy Cook for guests Go to the launderette Paint Go by bicycle Garden work
Tiredness Observed mean score
True IADL-score
9.770 7.510 5.670 4.040 3.440 2.890 1.770 2.010 1.710 1.340 0.790 0.190 0.110 0.190
9.162 4.464 3.370 3.166 2.618 2.124 1.019 1.374 1.016 0.845 0.448 0.065 0.065 0.033
376 Aging Clin. Exp. Res., Vol. 5, No.5
Make coffe Use pub!. transportation Cut bread etc. Wash clothes by hand Prepare cold meals Use the wash. machine Open cans/etc. Sew on the machine Manage economy Go to launderette Go by bicycle Cook for guests Paint Garden work
Observed mean score
True IADL-score
7.590 3.850 2.180 2.030 2.360 1.340 1.160 1.140 0.710 0.350 0.160 0.160 0.160 0.100
12.598 3.820 2.649 2.423 2.341 1.442 1.204 1.178 0.704 0.269 0.159 0.101 0.101 0.056
The measurement of Instrumental ADL
Table 4 - P-values for reduced functional ability on the two IADL-scales by various indicators of health problems. Men
lAD L-tiredness Women
IADL-reduced speed Men Women
Chronic disease Bronchitis Angina Pectoris Claudication Intermittens Osteoarthrosis
0.000 0.000 0.011 0.032
0.004 0.118 0.150 0.480
0.007 0.002 0.000 0.002
0.003 0.041 0.096 0.007
General symptoms Dyspnea Dizziness Walking-impairment Tiredness
0.000 0.003 0.000 0.000
0.000 0.000 0.014 0.000
0.000 0.000 0.000 0.004
0.000 0.003 0.036 0.000
Utilization of social and health services Hospitalization Technical aids Homehelp
0.000 0.000 0.000
0.009 0.016 0.001
0.485 0.000 0.004
0.828 0.020 0.001
X2 tests for difference in IADL among men and women in relation to the included health indicators.
The frequency of persons with reduced speed in all activities (score 0) is the same among men and women, whereas the share that gets tired in all activities is a little larger in men than in women. Generally more have problems with the reduced speed-scale than with the tiredness-scale. Table 5 - Distribution of men and women on the two IADL-scales. Per cent. IADL-tiredness Men Women Max score: Score 13: Score 12: Score 11: Score 10: Score 9: Score 8: Score 7: Score 6: Score 5: Score 4: Score 3: Score 2: Score 1: Score 0:
%
%
59.1 0.9 9.5 5.2 7.5 4.3 3.7 0.3 0.6 1.4 0.9 0.0 0.3 0.6 5.8
30.4 1.6 14.1 13.0 10.3 8.2 4.9 4.6 3.0 2.4 2.7 1.1 1.6 0.5 1.4
IADL-reduced speed Men Women %
%
41.2 1.4 11.5 8.6 9.8 6.6 2.0 0.6 1.7 2.3 0.6 0.9 0.3 1.4 11.0
26.9 2.7 8.7 12.2 12.8 6.3 3.8 8.9 8.9 6.3 8.9 1.3 2.5 6.3 10.1
The omitted items In the statistical analysis, 5 items fit poorly with the other items in the two scales; these were "shop", "prepare hot meals", "clean the floors", "needlework" and "knit", and 11 items only fit into one of the scales. Obviously, the number of omitted items is considerable. However, as the two IADL-scales in any case cover the area broadly, we chose to keep only those items accepted by the item analysis, and for the same reason keep only those items that are included in both the tiredness- and the reduced speed scale. Some of the omitted items deserve special comment. "Vacuum cleaning", "shopping", and "cooking" are items, which are included in most IADLmeasures. From a clinical point of view, these activities are of course important. "Vacuum cleaning" is an activity in the home, which is physically strenuous. However, activities in the home are already covered by eight items, and physically strenuous activities by four. "Shopping" is an activity outside the home which can also be physically strenuous. Activities "in town" are already described by four items. "Cooking" is covered by five other items. We, therefore, accept the omission of "vacuum cleaning" "shopping", and "preparing hot meals".
Aging Clin. Exp. Res., Vol. 5, No.5 377
K. Avlund, K. Schultz-Larsen, and S. Kreiner
DISCUSSION
This study describes a new measure of IADL that is able to discriminate among the large group of elderly who do not depend on help. The Rasch item analysis of IADL showed that 14 items can be combined into two qualitatively different additive scales; one includes 14 items defined in relation to reduced speed, and the other comprises the same 14 items, defined in relation to tiredness (Table 3). These two scales have an acceptable degree of validity, from a statistical point of view. In assessing whether the scales are acceptable from a more clinical/practical point of view, it is of importance to discern whether they cover a broad selection of daily functions. Some of the included activities take place in the home (prepare cold meals, cook for guests, make coffee/tea, cut bread, etc., open cans, wash clothes by hand, use the washing machine, paint, and manage economy), some around the home (paint, garden work) or outside the home (bicycle, use public transportation, manage economy, and go to the launderette). Some of the activities are physically strenuous (bicycle, garden work, paint, go to the launderette), and some are finer hand activities (cut bread/etc., open cans). Thus, the number of included IADL-items covers so many areas in the life of a person, that the extent to which a person is able to live a satisfying, independent and outgoing life can be actually measured by evaluating these 14 activities. The categories The category levels used in this study were especially developed in order to achieve a larger variability and a larger capacity for discriminating among a healthy elderly population. The rationale for using the categories of tiredness and reduced speed was developed earlier (26). The ability to maintain adequate speed when carrying out daily tasks contributes to a comfortable and independent life style in old age. The person who requires two hours to vacuum clean the livingroom has a considerable functional reduction in performing that activity. The time required by elderly persons to accomplish tasks that constitute daily activities was previously shown to be a useful indicator of functional dependency among elderly women aged 60 and older (33), among a group of institu-
378 Aging Clin. Exp. Res., Vol. 5, No.5
tionalized elderly persons (34), and among 79year olds in the community (35). The dimension of tiredness is of great importance in an elderly person's everyday life. If a person gets very tired after carrying out some daily tasks or after a long walk, he/she may stop doing it.
The tiredness aspect has not been widely used in the measurement of functional ability. However, it was included in the Nottingham Health Profile (36). Capacity to discriminate Many studies have shown that IADL scales discriminate better among the well elderly than do PADL scales (5-7,37-44). When describing daily activities in relation to tiredness and reduced speed, the measures of Physical and Instrumental ADL are both good discriminators among the community dwelling elderly. When using IADL to measure dysfunction in a community's elderly population, the numbers categorized as disabled differ. Among 65-75year olds, one study showed that 80% were independent in IADL (5) while another reported 87% (45). Among 70-74-year olds, Dawson et al. showed that 83% were independent in IADL (37), Fillenbaum found that about 71% were independent in IADL (4), Koyano et al. 94% (7) and Holstein et al. 76% (38). The IADL measure in this study is better able to discriminate among well elderly persons than earlier IADL-measures. Here 59.1% of men and 30.1 % of women manage IADL without tiredness, and 41.2% of men and 26.9% of women manage IADL without reduced speed. Furthermore, the measure distingUishes between elderly with and without specific chronic diseases and general symptoms, and between elderly users and non-users of health and social services. Therefore, this IADL measure has a large capacity for discriminating among the 70-year olds. Gender differences Generally, IADL measurement is problematic, because many IADL-activities are only performed by men or by women. By means of the Rasch item analysis, two scales were developed in which items were omitted when there were signs of sex-bias. These two IADL-scales can be used
The measurement of Instrumental ADL
to measuring IADL in both men and women, thus ensuring that any differences in the ability of the 70-year olds to perform the IADL-activities are due to actual differences in functional ability and not in gender. This study shows that elderly women are more disabled than men in relation to IADL, in agreement with findings of several other workers (4, 7, 24, 25, 37, 38, 46-48). However, with regard to the ability to perform PADL in the same population, no differences were found between sexes (26). This indicates that this measure of IADL is qualitatively different from the measure of PADL, and that it might be a more sensitive detector of early signs of ageing which presents in different ways among men and women. The gender differences in IADL may be due to the fact that causes of disability tend to be different among men and women. Manton (49) suggested that the reasons for disability among women often are less lethal than among men. Our study indicates that these sex differences are better detected by the measure of IADL than the measure of PADL. Criteria revisited The criteria set up in this study to measure IADL were met in the following way: the IADL measure covered a broad range of activities relevant to elderly people, and the ability to perform these activities was measured by classification categories that are relevant to elderly people; it distinguishes between what the elderly actually do and what they are capable of doing; it is a good discriminator among the group of elderly who do not depend on help, and it is possible to add the items in a valid way. However, to obtain valid IADL-indexes with a substantial discriminating power for both men and women, we omitted items that are highly relevant to especially elderly women, such as houseworkitems. A study of elderly Americans disclosed that what concerned them most in managing on their own was not being able to continue household activities (50). Other studies have shown problems with the reliability of these housework items (25, 51-58), and with the construct validity when trying to include them in a combined IADL index (53, 59-64). It follows that the criteria set up for the IADL-
measures described here are in some ways contradictory. This leads to several possible solutions: 1) the two scales described in this study with similar activities can be used in relation to either tiredness or reduced speed. This solution has high construct validity, but problems with content validity (sex-bias); 2) the two scales described with different activities can be used in relation to tiredness and to reduced speed (Table 2). This solution has a high construct validity, and the problems with sexbias are smaller as fewer "female" activities are omitted from the scales. These two solutions still make it possible to study the influence of other variables on IADL-function among both men and women; 3) different IADL-scales for men and women can be made. This way the construct validity can be kept, and the problems with sex-bias avoided, thus acknowledging that IADL is not the same among men and women. However, this solution produces difficulties when IADL-function among men and women are related to other variables in a study. This solution was previously proposed by Lawton and Brody in 1969 (40). If a common IADL-scale for men and women is desired, and a high degree of discriminatory power is demanded, we suggest using the measure described here as a supplement to existing dependency-scales of IADL. A questionnaire concerning IADL in the general population should then include the items presented in Appendix 2. Appendix 3 shows how the two IADL scales are formed. It should be stressed, however, that choice of ADL-scales depends on the purpose of the survey in which it will be employed. Final remarks The measure described in this study represents a new approach to the measurement of IADL among the healthy elderly. It obtains important information about the consequences of age-associated decline in functional ability in ways not previously used, and in areas previously not properly examined. It uses a scoring system that defines difficulty in terms of both tiredness and reduced speed; this offers a more sensitive scale than other measures of aging
Aging Clin. Exp. Res., Vol. 5, No.5 379
K. Avlund, K. Schultz-Larsen, and S. Kreiner
among healthy elderly people, especially with regard to differences in aging among men and women. In this way, it should be able to reflect functional status changes among the huge group of elderly that are not dependent on help, and consequently identify the effects of intervention or prevention. ACKNOWLEDGEMENTS This study has been supported by The Research Foundation of the Occupational Therapists' Union, The Medical Research Foundation of the Hospitals in the Counties of Bornholm, Frederiksborg, Roskilde, Storstmm and VestsjiEliand, The Danish Medical Research Council (SLF 126409), The Foundation of 1870, The Velux Foundation of 1981, and The Health Services Development fund.
REFERENCES 1. Fillenbaum G.G.: The well-being of the Elderly. Approaches to multidimensional assessment. World Health Organization, Geneva, 1984. 2. Kane RA, Kane RL.: Assessing the Elderly. A practical guide to measurement. Lexington Books, Toronto, 1981. 3. Feinstein AR, Josephy B.R., Wells C.K.: Scientific and clinical problems in indexes of functional disability. Ann. Intern. Med. 105: 413-420, 1986. 4. Fillenbaum G.G.: Screening the elderly. A brief instrumental Activities of Daily Living measure. J. Am. Geriatr. Soc. 33: 698-706, 1985. 5. Spector W.o., Katz S., Murphy J.B., Fulton J.P.: The hierarchical relationship between Activities of Daily Living and Instrumental Activities of Daily Living. J. Chronic Dis. 40: 481-489, 1987. 6. Asberg K.H., Sonn u.: The cumulative structure of Personal and Instrumental ADL. A Study of Elderly People in a Health Service District. Scand. J. Rehabil. Med. 21: 171-177, 1988. 7. Koyano W., Shibata H., Hakazato K., Haga H., Suyama Y., Matsuzaki T.: Prevalence of disability in Instrumental Activities of Daily Living among elderly Japanese. J. Gerontol. 43: S41-S45, 1988. 8. Katz S., Ford AB., Moskowitz RW., Jackson B.A, Jaffe MW.: Studies of illness in the aged. J.A.M.A. 12: 914-919, 1963. 9. Mahoney F.I., Barthel D.W.: Functional evaluation: The Barthel Index. M. State Med. J. 14: 61-65, 1965. 10. Ware J.E., Brook R.H., Davies AK, Lohr KN.: Choosing measures of health status for individuals in general populations. Am. J. Public Health 71: 620625, 1981.
380 Aging Clin. Exp. Res., Vol. 5, No.5
11. McDowell I., Newell c.: Measuring health: A gUide to rating scales and questionnaires. Oxford University Press, Oxford, 1987. 12. Kirshner 8., Guyatt G.: A methodological framework for assessing health indices. J. Chronic Dis. 38: 27-36, 1985. 13. Ware J.E. Jr.: Standards for validating health measures: Definition and content. J. Chronic Dis. 40: 473-480, 1987. 14. Guyatt G.H., Deyo R.A., Charlsen M., Levine M.N., Mitchell A: Responsiveness and validity in health status measurement: A clarification. J. Clin. Epidemiol. 42: 403-408, 1989. 15. Lawton M.P., Moss M., Fulcomer M., Kleban M.H.: A research and service oriented multilevel assessment instrument. J. Gerontol. 37: 91-99, 1982. 16. Fillenbaum G.G., Smyer M.A: The development, validity and reliability of the OARS Multidimensional Functional Questionnaire. J. Gerontol. 36: 428-434, 1981. 17. Hellevik 0.: Forskningsmetode i sosiologi og statsvitenskap, ed. 3. Universitetsforiaget, Oslo, 1984, pp. 108-140. 18. Andrich D.: Rasch models for measurement. Sage University Paper. Series on Quantitative Applications in the Social Sciences. Series no. 07-001. Sage Publications, Beverly Hills and London, 1988. 19. Cronbach L.J.: Test validation. In: Thorndike KL. (Ed.), Educational Measurement. American Council on Education, Washington D.C., 1971, pp. 443-507. 20. Foldspang A, Juul S., Olsen J., Sabroe S.: Epidemiologi. Sygdom og befolkning, ed. 2. Munksgard, Copenhagen, 1986. 21. Teresi J.A., Golden KR, Gurland B.J., Wilder D.E., Bennet KG.: Construct validity of indicator-scales developed from the Comprehensive Assessment and Referral Evaluation Interview Schedule. J. Gerontol. 39: 147-157,1984. 22. Avlund K, Schultz-Larsen K: What do 70-year-old men and women do? And what are they able to do? From the Glostrup Survey in 1984. Aging Clin. Exp. Res. 3: 39-49, 1991. 23. Fillenbaum G.G.: Assessment of health and functional status: An international comparison. In: Kane RL., Evans J.G., MacFayden D. (Eds.), Improving the health of older people. Oxford University Press, Oxford, 1990, pp. 69-90. 24. Heikkinen E., Waters W.E., Brzezinski Z.J.: The elderly in eleven countries. A sociomedical survey. World Health Organization, Copenhagen, 1983. 25. Waters W.E., Heikkinen E., Dontas AS.: Health, lifestyles and services for the elderly. Public Health in Europe. World Health Organization, Copenhagen, 1989.
The measurement of Instrumental ADL
26. Avlund K, Kreiner S., Schultz-Larsen K: Construct validation and the Rasch model: Functional ability of healthy elderly people. Scand J. Soc. Med. (In press). 27. Schultz-Larsen K, Avlund K, Kreiner S.: Functional ability of community dwelling elderly. Criterion validity of a new measure of functional ability. J. Clin. Epidemiol. 45: 1315-1326, 1992. 28. Hambleton RK, Rogers H.J.: Solving criterion-referenced measurement problems with item response models. Int. J. Educ. Res. 13: 145-160, 1989. 29. Rosenbaum P.: Criterion-related construct validity. Psychometrika 54: 625-634, 1989. 30. Lord F.: Applications of item response theory to practical testing problems. Lawrence Erlbaum Associates, Inc., New Jersey, 1980. 31. Guttman L.: The basis of scalogram analysis. In: Stouffer S.A, Guttman L., Suchman E.A, Lazarsfield P.F., Star S.A, Clausen J.A (Ed.), Measurement and prediction. Princeton University Press, New York, 1950, pp.60-90. 32. Cook L.L., Eignor D.R.: Using item response theory in test-score equating. Int. J. Educ. Res. 13: 161-173, 1989. 33. Williams M.E., Hadler N.M., Earp JAL.: Manual ability as a marker of dependency in geriatric women. J. Chronic Dis. 35: 115-122, 1982. 34. Williams ME, Hornberger J.e.: A quantitative method of identifying older persons at risk for increasing Long Term Care Services. J. Chronic Dis. 9/10: 705-711, 1984. 35. Lundgren-Lindquist B.: Functional capability in 79year-old women and men. University of Gothenburg; The Department of Rehabilitation Medicine and Geriatric and Long-Term Care Medicine, Thesis, 1982. 36. Hunt S.M., McKenna S.P., McEwen J., Williams J., Papp E.: The Nottingham Health Profile: Subjective health status and medical consultations. Soc. Sci. Med. 15A: 221-229, 1981. 37. Dawson D., Hendershot G., Fulton J.: Aging in the eighties. Functional limitations of individuals age 65 years and over. National Center for Health Statistics. Advance Data from Vital and Health Statistics. No. 133. DHHS Pub. No. (PHS) 87-1250. Public Health Service. Hyattsville, MD, June 10. 1987, pp. 1-12. 38. Holstein B., Almind G., Due P., Holst E.: fEldres selvrapporterede helbred og lCEgemiddelforbrug. Ugeskr. Laeger 152: 286-391, 1990. 39. Rosow I., Breslau N.: A Guttman health scale for the aged. J. Gerontol. 21: 556-559, 1966. 40. Lawton M.P., Brody E.M.: Assessment of older people: Self-maintaining and Instrumental Activities of Daily living. Gerontologist 9: 179-186, 1969. 41. Bergstrom G., Aniansson A, Bjelle A, Grimby G., Lundgren-Lindquist B., Svanborg A: Functional con-
sequences of joint impairment at age 79. Scand. J. Rehabil. Med. 17: 183-190, 1985. 42. Hendriksen C., Lund E., Str0mgard E.: fEldre menneskers funktionsformaen. En opg0relse fra "R0dovreprojektet". Ugeskr. Laeger 149: 1835-1839, 1987. 43. Manton KG.: The linkage of health status changes and disability. Comprehensive Gerontology A 1: 16-24, 1987. 44. Lammi U.K., Kivela S.K., Nissinen A, Pekkanen J., Punsar S.: Functional capacity and associated factors in elderly Finnish men. Scand. J. Soc. Med. 17: 67-75, 1989. 45. Manton KG.: A longitudinal study of functional change and mortality in the United States. J. Gerontol. 54: 625-634, 1988. . 46. Jette AM., Branch L.G.: The Framingham Disability Study: II. Physical disability among the aging. Am. J. Public Health 71: 1211-1216, 1981. 47. Gossman-Hedstrom G., Aniansson A, Persson G.B.: ADL-reduction and need for technical aids among 70year-olds. Comprehensive Gerontology B 2: 16-23, 1988. 48. Jette AM., Pinsky J.L., Branch L.G., Wolf PA, Feinleib M.: The Framingham Disability Study: Physical disability among community-clwelling survivors of stroke. J. Clin. Epidemiol. 41: 719-726, 1988. 49. Manton KG.: Epidemiological, demographic, and social correlates of disability among the elderly. Milbank Q. 67: 13-58, 1989. 50. Warren H.H.: Self-perception of independence among urban elderly. Am. J. Occ. Ther. 28: 329-336, 1974. 51. Wade D.T., Legh-Smith J., Hewer RL.: Social activities after stroke: measurement and natural history using the Frenchay Activities Index. Int. Rehabil. Med. 7: 176-181, 1985. 52. Kivela S.K.: Measuring disability - do self-ratings and service provider ratings compare? J. Chronic Dis. 37: 115-123, 1984. 53. Jette AM., Deniston O.L.: Inter-observer reliability of a functional status assessment instrument. J. Chronic Dis. 31: 573-580, 1978. 54. Pollard WE, Bobbitt RA, Bergner M., Martin D.P., Gilson B.S.: The Sickness Impact Profile: Reliability of a health status measure. Med. Care 14: 146-155, 1976. 55. Sheikh K, Smith D.S., Meade TW., Goldenberg E., Brennan P.J., Kinsella G.: Repeatability and validity of a modified Activities of Daily Living (ADL) index in studies of chronic disability. Int. Rehab. Med. 1: 51-58, 1979. 56. Haworth R.K, Hollings E.M.: Are hospital assessments of daily living activities valid? Int. Rehab. Med. 1: 59-62, 1979. 57. Parkerson G.R, Gehlbach S.H., Wagner E.H., James SA, Clapp N.E., Muhlbaier L.H.: The Duke-UNC
Aging Clin. Exp. Res., Vol. 5, No.5 381
K. Avlund, K. Schultz-Larsen, and S. Kreiner
Health Profile: An adult health status instrument for primary care. Med. Care 19: 806-828, 1981. 58. Sonn U., Asberg K.H.: Assessment of Activities of Daily Living in the elderly. A study of a population of 76-year-olds in Gothenburg, Sweden. Scand. J. Rehabil. Med. 23: 193-202, 1991. 59. Pfeffer R.I., Kurosaki TT, Harrah c.H., Chance J.M., Rlos S.: Measurement of functional activities in older adults in the community. J. Gerontol. 37: 323-329, 1982. 60. Teresi J.A., Cross P.S., Golden RR: Some applications of latent trait analysis to the measurement of ADL. J. Gerontol. 44: S196-S204, 1989. 61. Fugl-Meyer A.R., Jaasko L.: Post-stroke hemiplegia and ADL-performance. Scand. J. Rehabil. Med. (Suppl 7): 140-152,1980. 62. Rubenstein L.Z., Schairer c., Wieland GD., Kane R: Systematic biases in functional status assessment of elderly adults: Effects of different data sources. J. Gerontal. 39: 686-691, 1984. 63. Magaziner J., Simonsick E.M., Kashner TM., Hebel J.R: Patient-proxy response comparability on measures of patient health and functional status. J. Clin. Epidemiol. 41: 1065-1074, 1988. 64. Lindmark B., Hamrin E.: Instrumental Activities of Daily Living in two patient populations, three months and ,one year after stroke. Scand. J. Caring Sci. 3: 161-169,1989. 65. Andersen E.B.: Discrete statistical models with social science applications. North-Holland Publishing Company, Amsterdam, New York, Oxford, 1980. 66. Gustafsson J-E.: Testing and obtaining fit of data to the Rasch Model. Br. J. Math. Stat. Psychol. 33: 205233, 1980. 67. Kreiner S.: Analysis of multidimensional contingency tables by exact conditional tests: Techniques and strategies. Scandinavian Journal of Statistics 14: 97-112, 1987. 68. Kreiner S.: Rasch 1 & 2. Programs for item analysis. Technical Report. Danish Institute for Educational Research, Copenhagen, 1985.
Appendix 1: Techniques for Rasch item analysis Tests and techniques include (65, 66): 1) Global tests for homogeneity across scoregroups. Internal homogeneity. Andersen's (65) conditional likelihood ratio test, Z, based on the conditional distribution of responses given the total score is the appropriate statistic here. 2) Global tests for homogeneity across groups defined by external criteria (Gender, Age, Social Economic Status). External homogeneity. Andersen's Z-statistic once again. 3) Analysis of item response curves and item-by-score
382 Aging Clin. Exp. Res., Vol. 5, No.5
residuals including likelihood ratio tests for homogeneity of separate items. 4) Tests for local independence of items by analysis of interaction between separate items conditionally given the total score. (The item-by-item two-way tables must display interactions for items even in separate scoregroups. We are, however, able to estimate the degree to which interaction must be present. Deviance from this hypothetical interaction is evidence against local independence). 5) Tests for homogeneity across subscales. Per MartinLof's likelihood ratio test (66) is appropriate here. This test assumes that each of two subscales may be described by a Rasch model and tests whether or not the underlying item-structures correspond to the structure estimated for the complete scale. 6) Tests against manifest item-bias with respect to external variables of interest. Test for item bias are very simple tests for conditional independence in three-way contingency tables, but nevertheless surprisingly informative. Exact conditional tests and tests for ordinal categorical variables may easily be applied here if necessary (67). 7) Finally, various graphical tests and techniques have been suggested, the most important probably being scatter plots of item-marginals for specific sub-populations. Items with low scores in one sub-populations must be the items with low scores in all other sub-populations for the Rasch model to hold. These plots should therefore always show a strong positive correlation. Specialized software for these procedures have been developed (68), but most calculations can be done using standard statistical programs like GUM and GENSTAT.
Appendix 2: Questionnaire of IADL 1. Are you able to bicycle? 1. yes 2. no 3. not relevant yes no If yes: a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2 2. Are you able to use public transportation? 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2 3. Are you able to administrate your economy? 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2 4. Are you able to prepare cold meals? 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2
The measurement of Instrumental ADL
S. Are you able to make coffee/thea 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2
12. Are you able to use the sewing machine? 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2
6. Are you able to cut bread/meat/vegetables? 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2
13. Are you able to paint in and around the house? 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2
7. Are you able to open cans/etc? 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2
14. Are you able to work in the garden? 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2
8. Are you able to cook for guests? 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2 9. Are you able to wash clothes by hand? 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2 10. Are you able to use the washing machine? 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2 11. Are you able to go to the launderette? 1. yes 2. no 3. not relevant If yes: yes no a) Do you get tired? 1 2 b) Does it take more time than earlier? 1 2
Appendix 3: The IADL-scales The answers to the questions in the questionnaire in Appendix 2 are combined into two IADL-scales which are scored from the following questions: IADL-tiredness: la, 2a, 3a, 4a, Sa, 6a, 7a, 8a, 9a, lOa, 11a, 12a, 13a, 14a. IADL-reduced speed: 1b, 2b, 3b, 4b, Sb, 6b, 7b, 8b, 9b, lab, lIb, 12b, 13b, 14b. The two IADL-scales are calculated the same way. The individual items are coded one or zero. The scales count the number of items coded 1, that is the number of items, where the function is not reduced. High value thus describes better function, low scale value describes worse function.
Aging Clin. Exp. Res., Vol. S, No. S 383