Research Article

Losing the Ability in Activities of Daily Living in the Oldest Old: A Hierarchic Disability Scale from the Newcastle 85+ Study

  • Andrew Kingston mail, (AK); (CJ)

    Affiliation: Institute for Ageing and Health, Newcastle University, Newcastle upon Tyne, United Kingdom

  • Joanna Collerton,

    Affiliation: Institute for Ageing and Health, Newcastle University, Newcastle upon Tyne, United Kingdom

  • Karen Davies,

    Affiliation: Institute for Ageing and Health, Newcastle University, Newcastle upon Tyne, United Kingdom

  • John Bond,

    Affiliations: Institute for Ageing and Health, Newcastle University, Newcastle upon Tyne, United Kingdom, Institute for Health and Society, Newcastle University, Newcastle upon Tyne, United Kingdom

  • Louise Robinson,

    Affiliations: Institute for Ageing and Health, Newcastle University, Newcastle upon Tyne, United Kingdom, Institute for Health and Society, Newcastle University, Newcastle upon Tyne, United Kingdom

  • Carol Jagger mail (AK); (CJ)

    Affiliation: Institute for Ageing and Health, Newcastle University, Newcastle upon Tyne, United Kingdom

  • Published: February 15, 2012
  • DOI: 10.1371/journal.pone.0031665



To investigate the order in which 85 year olds develop difficulty in performing a wide range of daily activities covering basic personal care, household care and mobility.


Cross-sectional analysis of baseline data from a cohort study.


Newcastle upon Tyne and North Tyneside, UK.


Individuals born in 1921, registered with participating general practices.


Detailed health assessment including 17 activities of daily living related to basic personal care, household care and mobility. Questions were of the form ‘Can you …’ rather than ‘Do you…’ Principal Component Analysis (PCA) was used to confirm a single underlying dimension for the items and Mokken Scaling was used to determine a subsequent hierarchy. Validity of the hierarchical scale was assessed by its associations with known predictors of disability.


839 people within the Newcastle 85+ study for whom complete information was available on self-reported Activities of Daily Living (ADL). PCA confirmed a single underlying dimension; Mokken scaling confirmed a hierarchic scale where ‘Cutting toenails’ was the first item with which participants had difficulty and ‘feeding’ the last. The ordering of loss differed between men and women. Difficulty with ‘shopping’ and ‘heavy housework’ were reported earlier by women whilst men reported ‘walking 400 yards’ earlier. Items formed clusters corresponding to strength, balance, lower and upper body involvement and domains specifically required for balance and upper/lower limb functional integrity.


This comprehensive investigation of ordering of ability in activities in 85 year olds will inform researchers and practitioners assessing older people for onset of disability and subsequent care needs.


Activities that are required to function independently in daily life, so called activities of daily living (ADLs), have long been seen as essential measures of disability in ageing studies and in clinical practice to assess care needs. When Basic (personal care) Activities of Daily Living (BADLs), for instance feeding, bathing, and toileting [1], are combined with Instrumental Activities of Daily Living (IADLs) which measure the ability to self-care within a household through activities such as shopping, cooking and doing housework [2], they better describe the spectrum of disability for a broader range of people [3]. In addition some researchers discriminate between BADLs, IADLs and mobility items such as walking a short distance, using steps and moving around the home, the latter comprising functional limitations (specific actions) rather than compound actions that form activities.

The hierarchical structure to the order in which loss of ability in both BADL and IADL items occurs has been confirmed by cross-sectional and longitudinal studies [3][7]. When both BADL and IADL items are considered together, difficulty with IADL items precedes difficulty with BADL items within the hierarchy [3]. The order of loss of ability to perform activities has also been classified in terms of four domains with each domain containing multiple activities that are similar in terms of their need for specific functional integrity combinations of dexterity, balance, strength and upper or lower extremity involvement [5]. For example the first abilities lost require manual dexterity and the last upper rather than lower limb control.

A number of issues remain unresolved in terms of the hierarchy of activities. Most studies have included only a small set of items, typically five or six and, if not selected to span the full range of disability, may result in floor effects. Few studies have investigated hierarchies separately for men and women, particularly important for IADLs such as cooking, which may be confounded with sex-specific household roles, so-called situational disability as opposed to ‘true’ functional disability [8]. In addition few studies have included large numbers of the oldest old, those aged 85 years and older, who represent the fastest growing section of the population [9].

The aims of this paper were to develop a hierarchical disability scale, using a much wider range of BADL, IADL and mobility items than previously employed which was appropriate for the men and women using cross-sectional (baseline) data from a birth cohort of over 800 85 year old participants in the Newcastle 85+ Study [10], [11]. In addition we aimed to validate the scale by examining its relationship with known predictors of disability.

We feel that further confirmation of the ordering of loss of BADLs and IADLs in this unselected single year birth cohort across a much wider range of activities, would assist researchers and clinicians in choosing subsets of activities that span the whole spectrum of disability and deepen understanding of the order in which older people lose functional capacity, thereby facilitating the design of more appropriate aids and appliances and the targeting of resources during the onset and progression of disability.


The study population for analysis comprised 839 of the health assessment sample (98.8%) this being participants who had both a health assessment with complete data on all BADL, IADL and mobility items, and a review of general practice records. Table 1 shows that women formed 62% (520) of the study population; 10.1% (85) of the study population resided in an institution and women were at least twice as likely to live in an institution compared with men (p = 0.004,OR = 2.1), 7.0% (58) had severe cognitive impairment (SMMSE 0–17) with no difference in prevalence between men and women (p = 0.54), and 25.5% (211) had three or more long standing illnesses, again with no difference between men and women (p = 0.15). Women were nearly twice as likely to take more than 12 seconds to complete the timed up-and-go compared to men (P<0.001, OR = 1.9). There was no statistically significant difference in the number of prescribed medications between men and women (p = 0.07) or in the number of diagnosed diseases they had (p = 0.10).


Table 1. Basic Descriptive Statistics of the Study Population by Sex.


Participants experienced most difficulty with cutting toenails, shopping and using steps and least with washing hands and face and feeding (Table 2). Women experienced significantly more difficulty than men with all items except dressing and light housework.


Table 2. Prevalence of ‘Difficulty’ in (I)ADL and Mobility Items - %(n).


PCA identified for men that one component (eigenvalue = 12.8) explained 75.4% of the variance with the second component (eigenvalue = 1.1) capturing only 6.7% further. Similar results were found for women, with the first component (eigenvalue = 11.0) capturing 71.8% of the variance and the second component (eigenvaue = 1.2) explaining a further 7.2%. This was further confirmed when men and women were analysed together; component 1 accounted for 76.2% of variation (eigenvalue 11.9); component 2 explained 6.2% of the variation (eigenvalue = 1.1). Whether men and women were analysed together or separately we found strong evidence of unidimensionality with component one always producing a far greater eigenvalue than component two and with approximate equal loading factors across all items (factor loadings all: minimum = 0.188, maximum = 0.262; men: minimum = 0.186, maximum = 0.271; women: minimum = 0.192, maximum = 0.268). However when men and women were analysed together, the second component indicated larger loading factors for the ‘manage money’ and ‘manage medication’ items, suggesting that this could be a dimension related to cognition, although this was less evident for men alone.

Mokken Scaling indicated that a hierarchy was present within the data and confirmed the unidimensionality conclusions of the PCA (Loevinger Scalability Co-efficient = 0.68). All items satisfied the assumption of single monotonicity thus suggesting that each item forms at distinct loci on a disability scale i.e. no items were measuring disability at exactly the same level. However, five items violated the assumption of double monotonicity: ‘transfer from chair’, ‘transfer from toilet’, ‘manage medications’, ‘move around the home’ and ‘manage money’ (women only) (Table 3). As the PCA indicated a possible second dimension related to cognition, loading on ‘managing money’ and ‘managing medications’, and since these items also failed the assumption of double monotonicity when men and women were analysed separately, the Mokken Scaling was repeated with these items removed. Removal of these items increased the strength of the hierarchical scale (Loevinger Scalability Coefficient ≥0.71) (Table 1), removed any further violation of assumptions and was significantly better than when cognition items were included. All analyses reported subsequently were calculated with the cognition items removed.


Table 3. Hierarchy of Loss of Ability in (I/B)ADL and Mobility Items Formed by Mokken Scaling.


For both men and women, ‘cutting toenails’ was the first activity with which participants had difficulty, and feeding was the last. The scaling algorithm constructed a numerical ordering for the items which indicated whether items had a tendency to be clustered together in terms of difficulty. When the scores of relative difficulty in performance of the items were plotted (Figure 1), clear areas of clustering of items was evident, corresponding with previously reported domains requiring particular combinations of lower and upper body strength combined with balance [5]; these being indicated in Figure 1 by; (A) – involving complex manual dexterity and balance; (B) – long distance mobility and balance; (C) – upper limb control and standing balance and (D) – upper limb control in a seated position.


Figure 1. Relative Difficulty of BADL, IADL and Mobility Items (Domain of Disability [5]).

Abbreviations: BADL – Basic Activities of Daily Living. IADL – Instrumental Activities of Daily Living.


The disability scale formed from assigning participants to the highest hierarchical position of the items with which a participant had difficulty, was highly correlated with the more usual scale formed by summing the number of items (out of 15) with which the participant had difficulty (Spearman's ρ = 0.94) and it had very strong internal consistency (Cronbach's α = 0.937). Further validation by comparison with known predictors of disability (Table 4) showed a significant association with all measures apart from education.


Table 4. Association of the Hierarchical Disability Scale with Known Predictors of Disability - %(n).


Exclusion of those residing in institutions had no effect on either the PCA or the Mokken Scaling procedure. Similarly, using a cut point of ‘needing help’ rather than ‘difficulty’ for the BADL, IADL and mobility items did not alter the conclusions.


The Newcastle 85+ Study [10], [11] recruited a cohort of 1040 85 year olds from general practices in Newcastle and North Tyneside, UK. Eligible individuals were all those born in 1921 (aged around 85 at the time of recruitment) and who were permanently registered with a general practice in the study area. 83% (53/64) of general practices agreed to take part; those who declined were similar on practice size, the proportion who were training practices, National Health Service (NHS) Quality and Outcomes Framework score Index of Multiple Deprivation score (IMD) for 2004 to those agreeing. Participating general practitioners were asked to review patient lists prior to mail-out and to exclude only those individuals with end stage terminal illness (n = 11). All those who met these inclusion criteria were invited to participate (n = 1459), whether living at home or in an institution, and regardless of their state of health with recruitment and assessment taking place over a 17 month period during 2006–2007. A total of 358 people (24.5%) declined to participate, these being similar in terms of sex and deprivation to those who agreed to take part [10]. Study participants were assessed in their normal place of residence, including institutional care, by trained research nurses with a series of questionnaires, measurements, function tests, a blood test and a review of general practice records. Participants could decline parts of the assessments. Of the potential sample of 1040 people, 849 agreed to the health assessment and a review of general practice records; 188 to GP record review only and 3 agreed only to take part in the health assessment. Fewer females were in the health assessment plus record review group (62.0%, 526/849) than in the record review only group (72.3%, 136/188) (full details of the design of the study can be found elsewhere [10], [11]).

The present analysis was confined to those participants who had the health assessment since this was the only source of information on ADLs. During the health assessment, participants were asked if they were able to do the following activities: cut toenails, wash all over, transfer from a bed/toilet/chair, dress and undress, wash face and hands and self-feed (including cutting up food), shop for groceries, do light housework, do heavy housework, manage money, manage medications and prepare and cook a hot meal. In addition participants were asked three questions on mobility: get around in the house, go up and down stairs/steps, and walk at least 400 yards? Each question was framed as ‘can you’ rather than ‘do you’ to have greater capacity to assess true levels of disability [12] accounting for situational responses. Responses to all items were: I have no difficulty doing this by myself/ I have some difficulty doing this by myself/ I can only do this by myself if I use an aid or appliance/ I am unable to do this by myself, I need someone's help.

Socio-demographic information included sex, years of education, and institutional status, with additional variables including the number of longstanding illnesses; the number of prescribed medications (extracted from GP records); a disease count from the presence of 18 selected chronic diseases [10] and the timed up and go test [13] with a cut point that determines those with normal function as performing the test in 12 seconds or less [14]. Cognitive function was also measured by the Standardised Mini-Mental State Examination [15], [16] with severe cognitive impairment classified by a score of 17 or less out of 30, this cut point having high sensitivity for moderate and severe dementia [17].

Ethnical approval was obtained from Newcastle & North Tyneside Local Research Ethics Committee One and informed written consent was obtained from all participants.

To determine whether the BADL, IADL and mobility items formed a single dimension we used Principal Component Analysis (PCA) based on the polychoric correlations between items and with whole ordinal scales. The number of dimensions was determined using Kaiser's Criterion, including only eigenvalues greater than one [18]. Having identified a single dimension, we dichotomised the items using a cut point of no difficulty/some difficulty (from a four category response of ‘no difficulty’, ‘some difficulty’, ‘only with and aid’ and ‘unable to do this’). We then used Mokken Scaling to verify the unidimensionality and to determine the hierarchy from the Loevinger Scalability Coefficient (H) [19] with values of H between 0.3–0.39 being taken to suggest a weak Mokken scale; between 0.4–0.49 an acceptable Mokken scale and greater than 0.5 a strong Mokken scale [20]. Items were deleted from the scale if they did not satisfy the assumption of single monotonicity (each item forms at a distinct loci on a scale of decreasing difficulty) and double monotonicity (the Item Characteristic Curves are non-overlapping). As Mokken scaling uses multiple tests on the data a Bonferroni correction was implemented to reduce the type I error. A scoring system was formed based on the highest item in the hierarchy with which the participant had difficulty (a score of 1 being low indicating difficulty with the first (most difficult) item in the hierarchy and a score of 17 being the highest indicating difficulty with then last (least difficult) item in the scale). Participants having no difficulty with all items in the scale were assigned a score of zero. Cronbach's alpha [21] was used to assess the internal consistency of the scale with values close to one suggesting a strong scale and values close to zero indicative of poor internal consistency. We constructed a disability scale corresponding to the highest hierarchical position of the items with which the participant had difficulty. After separation into four categories; difficulty with no items, difficulty with 1–5 items, difficulty with 6–10 items and difficulty with 11–15 items (to allow for nonlinear associations) the scale was validated against known predictors of disability [22].

Separate analyses were carried out for men and women. Sensitivity analyses were undertaken excluding the participants living in institutions and using the alternative cut-point for performance of no help required/help required. All analyses were carried out in Stata 10.1 [StataCorp. 2009. Statistical Software: Release 10.1. College Station, TX: Stata] with statistical significance at α = 0.05.


We found a strong hierarchical ordering to loss of ability in a wide range of basic and instrumental activities of daily living and the items measuring mobility in an unselected population aged 85 years in 2006. ‘Cutting toenails’ was the first item with which participants found difficulty and ‘washing hands and face’ and ‘feeding’ the last items. The ordering of the items in our hierarchic scale confirms previous studies using cross-sectional [3], [5], [6], [23] and longitudinal [4], [7] data, but which are now 10–20 years and were based on a more restricted set of items predominantly in the younger old. Thus our study adds considerably to the evidence that the order of loss of activities does not vary with age. Sex differences were evident from our single birth year cohort; not only were women more likely to report difficulty with each activity than were men but we also found that the ordering of loss differed between men and women of the same age with women reporting more difficulty with activities requiring strength (‘shopping’ and ‘heavy housework’) whilst men were earlier in reporting difficulty walking. Moreover the ordering and our conclusions were unchanged if inability to perform was defined as requiring the help of another person rather than the more unbiased having difficulty performing alone. However measuring disability by the requirement for ‘help’ may depend on the availability of help which may therefore bias results [24] and thus our primary measure based on ‘difficulty’ adds strength to our study.

Previous research has indicated that disability in later life appears to progress with difficulty in IADLs preceding that with BADLs [4], [6], [7], [23]. Though this was broadly true in our analysis; there was overlap in the ordering of IADLs and BADLs and our ordering was much more consistent with the domains of disability defined by Ferrucci et al [5] which combine IADL, BADL and mobility items requiring similar underlying impairments. They defined the first domain in which difficulty would be as activities requiring complex manual dexterity coupled with balance such as ‘cutting toenails’ and ‘heavy housework’. Activities in the next domain require balance and involve the capacity to walk long distances, the equivalent activities in our study being ‘shopping’, ‘use steps’, ‘walk 400 yards’, ‘full wash’, ‘cooking a hot meal’, ‘light housework’. The third domain in decreasing difficulty contained activities requiring standing balance and good upper limb control; ‘move around the home’, ‘transfer from chair’, ‘toilet’, ‘dressing’ and ‘transfer from bed’. The final domain, and the easiest to perform, related to good upper limb control when in a seated position; ‘washing face and hands’ and ‘feeding’. Only two items appeared out of step between our hierarchy and Ferrucci's four domains; ‘heavy housework’ and ‘light housework’ and it may be that the perceived meanings and the nature of these tasks have changed more over time than for other activities.

Our disability scale formed from the hierarchy performed well when examined alongside known predictors of disability. The ordering of loss of activities is of potential use to others selecting activities to measure a range of severity of disability both in the research and clinical setting. We collected information from over 800 older people aged 85 years of age on their situational disability (questions being framed as ‘can you’ rather than ‘do you’) from 17 IADLs, BADLs and mobility items, a much larger number than previous studies and with minimal missing data. The Newcastle 85+ Study has a broad range of health measures and is representative of the larger population of older people in Newcastle upon Tyne [10].

The main limitation of our study is the cross-sectional nature of the data although our results were in agreement with the previous longitudinal studies [4], [7] Nevertheless, unlike other studies, our population came from a single birth cohort with a high response rate and included those in institutions. Thus the sex difference we found in the order of loss of activities was not due to the greater average age of women compared to men in general older populations. Inclusion of those in institutions where there is a high prevalence of dementia may be viewed as a limitation. We had a small proportion of our sample (n = 71, 8.5%) who had a diagnosis of dementia and these participants may lose ability to perform activities independently in a different order to those without dementia, for instance they may be able to walk 400 yards but may not be able to dress themselves. However, we repeated all analyses excluding those in institutions and the conclusions were unchanged. Nevertheless, when we originally included items more dependent on complex cognitive ability than physical ability (‘managing money’ and ‘managing medication’) we found that they did not satisfy all of the underlying assumptions of the scaling method and their subsequent exclusion strengthened the scale formed although the remaining items may be viewed as being more ‘physical’.

Results of this analysis could therefore provide information to help identify older people at risk of functional decline and for the allocation, and prioritisation of, community services and social support to enable independent living for as long as possible. We found in terms of the ranking of difficulty that a number of items were ranked similarly suggesting that if these items alone were selected for inclusion in a disability scale, or indeed for assessment purposes, then the range of severity would be limited. ‘Cutting toenails’ and ‘shopping’ were the items with which our population most commonly reported difficulty. The former requires good balance and manual dexterity whilst the latter requires upper body strength and mobility. Physical activity programmes to delay the onset of disability should perhaps focus on exercises to improve these functions; in addition our results could support an argument for the essential core provision of specific services such as chiropody within the community care, as opposed to the current limited provision.


We are greatly indebted to the participants in the study for the generous donation of their time and personal information to make the study possible. We thank the research nurses (Brenda Balderson, Sally Barker, Julie Burrows, June Edwards, Julie Ferguson, Gill Hedley, Joan Hughes, Judith Hunt, Julie Kimber, and Victoria Raynor), data manager (Pauline Potts) and study secretary (Lucy Farfort) for outstanding work at all times. We acknowledge, with gratitude, the support of Newcastle and North Tyneside Primary Care Trusts and local general practices, and the advice given by the Newcastle and North Tyneside 1 Research Ethics Committee. We thank Rudi Westendorp and Jacobijn Gussekloo (Leiden University Medical Centre, the Netherlands) for generously sharing with us their experiences of the Leiden 85+ Study. We also thank Tom Kirkwood, Thomas von Zglinicki, Martin Eccles and Carmen Martin-Ruiz for helpful comments on drafts of the manuscript.

Author Contributions

Conceived and designed the experiments: AK JC KD JB LR CJ. Performed the experiments: AK JC KD JB LR CJ. Analyzed the data: AK CJ. Wrote the paper: AK JC KD JB LR CJ.


  1. 1. Katz S, Ford A, Moskowitz R, Jackson B, Jaffe M (1963) Studies of illness in the aged: the index of ADL, a standardised measure of biological and psychosocial function. JAMA 914–919.
  2. 2. Lawton M, Brody E (1969) Assessment of Older People: self maintaining and instrumental activities of dialy living. Gerontologist 9: 179–186.
  3. 3. Kempen GIJM (1995) Hierarchical Structure in ADL and IADL: Analytical Assumptions and Applications for Clinicians and Researchers. Journal of Clinical Epidemiology 48: 1299–1305.
  4. 4. Dunlop DD (1997) Disability in Activities of Daily-Living: Patterns of Change and a Hierarchy of Disability. American Journal of Public Health 87: 378–383.
  5. 5. Ferrucci L, Guralnik JM, Cecchi F, Marchionni N, Salani B, et al. (1998) Contstant Heirarchic Patterns of Physical Functioning Across Seven Populations in Five Countries. The Gerontologist 38: 286–294.
  6. 6. Weiss CO, Fried LP, Bandeen-Roche K (2007) Exploring the Hierarchy of Mobility Performance in High-Functioning Older Women. Journal of Gerontology 62A: 167–173.
  7. 7. Jagger C, Arthur AJ, Spiers NA, Clarke M (2001) Patterns of onset of disability in activities of daily living with age. Journal of the American Geriatrics Society 49: 404–409.
  8. 8. Deeg DJH (1993) Sex differences in IADL in the Netherlands: Functional and Situational disability. In: Robine J-M, Mathers CD, Bone MR, Romieu I, editors. Calculation of health expectancies; Harmonization, Consensus Achieved and Future perspectives. Paris: John Libby Eurotext.
  9. 9. United Nations Department of Economic and Social Affairs Population Division (2002) World Population Ageing: 1950–2050. New York: United Nations.
  10. 10. Collerton J, Davies K, Jagger C, Kingston A, Bond J, et al. (2009) Health and disease in 85 year olds: baseline findings from the Newcastle 85+ cohort study. BMJ 339: b4904.
  11. 11. Collerton J, Barrass K, Bond J, Eccles M, Jagger C, et al. (2007) The Newcastle 85+ study: biological, clinical and psychosocial factors associated with healthy ageing: study protocol. BMC Geriatrics 7: 14.
  12. 12. Glass TA (1998) Conjugating the “Tenses” of Function: Discordance Among Hypothetical, Experimental, and Enacted Function in Older Adults. The Gerontologist 38: 101–112.
  13. 13. Podsialdo D, Richardson S (1991) The timed “up & go”: a test of basic functional mobility for frail elderly persons. Journal of the American Geriatric Society 39: 142–148.
  14. 14. Bischoff HA, Stahelin HB, Monsch AU, Iversen MD, Weyh A, et al. (2003) Identifying a cut-off point for normal mobility: a comparison of the timed ‘up and go’ test in community-dwelling and institutionalised elderly women. Age Ageing 32: 315–320.
  15. 15. Molloy DW, Alemayehu E, Roberts R (1991) Reliability of a Standardized Mini-Mental State Examination compared with the traditional Mini-Mental State Examination. Am J Psychiatry 148: 102–105.
  16. 16. Molloy DW, Standish TIM (1997) A Guide to the Standardized Mini-Mental State Examination. International Psychogeriatrics 9: 87–94.
  17. 17. Jagger C, Clarke M, Anderson J (1992) Screening for dementia—a comparison of two tests using receiver operating characteristic (ROC) analysis. International Journal of Geriatric Psychiatry 7: 659–665.
  18. 18. Jolliffe IT (2002) Principal Component Analysis: Springer.
  19. 19. Diesfeldt HFA (2004) Executive functioning in psychogeriatric patients: scalability and construct validity of the Behavioral Dyscontrol Scale (BDS). International Journal of Geriatric Psychiatry 1065–1073.
  20. 20. Olsen LR, Mortensen EL, Bech P (2004) The SCL-90 and SCL-90R versions validated by item response models in a Danish community sample. Acta Psychiatrica Scandinavica 110: 225–229.
  21. 21. Bland JM, Altman DG (1997) Statistics notes: Cronbach's alpha. BMJ 314: 572-.
  22. 22. Stuck AE, Walthert JM, Nikolaus T, Büla CJ, Hohmann C, et al. (1999) Risk factors for functional status decline in community-living elderly people: a systematic literature review. Social Science & Medicine 48: 445–469.
  23. 23. Njegovan V, Man-Son-Hing M, Mitchell SL, Molnar FJ (2001) The Hierarchy of Functional Loss Associated With Cognitive Decline in Older Persons. J Gerontol A Biol Sci Med Sci 56: M638–643.
  24. 24. Verbrugge LM, Jette AM (1994) The disablement process. Social Science & Medicine 38: 1–14.