Abstract
Abstract
Objectives
This study examines national patterns of functional impairment, and how they vary by the presence of non-communicable disease (NCD), type of health condition, comorbidity, age, sex, ethnicity, deprivation and living situation.
Design
A cross-sectional examination using a national research database of linked administrative and survey data sets including census, tax and health data.
Setting
Aotearoa New Zealand
Participants
All individuals living in NZ on 30 June 2018, identified by the Statistics NZ Integrated Data Infrastructure estimated residential population (4.79 million individuals). Nine NCDs among these individuals were identified from national health data sets using existing and adapted algorithms.
Primary and secondary outcome measures
Functional impairment was assessed via hospitalisations, comorbidities (Elixhauser index), activity limitations (census) and income support.
Results
Three-tenths (29%) of the population had at least one NCD. Functional impairment was strongly patterned by NCD prevalence (3% activity limitation in people without any NCDs, 13% if one or more NCDs and 25% if two or more NCDs). Activity limitation was most common in individuals with dementia (68% activity limitation), stroke (42%) and coronary heart disease (26%). After age stratification, there was also a high level of activity limitation and income support in people with mental health conditions. Māori and Pacific peoples and people living in deprived areas or alone were more likely to have functional impairment.
Conclusions
Functional impairment was strongly patterned by NCD type. NCD prevention efforts and disability supports are needed to reduce the burden of disability experienced.
Keywords: Hospitalization, Chronic Disease, Dementia, Stroke, Coronary heart disease, Disabled Persons
STRENGTHS AND LIMITATIONS OF THIS STUDY.
National-level linked data sets enabled health information on non-communicable disease (NCD) diagnoses to be linked to census measures of functional impairment.
The whole population study had excellent power with nearly 5 million participants to explore patterns in the distribution of the disability burden by NCD and social factors.
The cross-sectional analysis examines associations between the burden of NCDs and functional impairment, but this type of analysis does not allow us to make causal inferences about the impact of NCDs on functional impairment.
Identifying existing NCDs relied on data availability, coverage over time and quality. Although we had no primary care data, we community prescriptions data were used.
Incomplete coverage of the 2018 census may have underestimated functional impairment in Māori, Pacific and high deprivation groups, causing our results to be conservative.
Introduction
Non-communicable diseases (NCDs) are the major contributor to premature death and disability globally and in Aotearoa New Zealand (NZ),1 and to inequalities in premature mortality.2 More than half of health spending in NZ has been attributed to NCDs, particularly cardiovascular, musculoskeletal, neurological diseases and cancer.3 At the same time, many NCDs are preventable, for example with reductions in tobacco, obesity and alcohol use, and improvements in diet and physical activity. NCDs are generally chronic/long-term health conditions. The impacts of NCDs are exacerbated in individuals experiencing multiple comorbid health conditions. For example, cancer patients with comorbidity have poorer survival, poorer quality of life and higher healthcare costs.4
Disability, which limits a person’s movement or activities, is intricately associated with NCDs.5 Functional impairment is an operational measure of disability.6 A systematic review of four chronic conditions showed that high proportions of people with cancer (10%–35%), cardiovascular disease (21%–64%), chronic respiratory disease (7%–50%) and diabetes (12%–55%) experienced difficulties with activities of daily living (ie, different proportions for eating, bathing and dressing for each NCD).5 There was a lot of variation in the level and types of disability by NCD type.5 For example, the review identified high levels of disability after a stroke, and mobility and visual impairments were common impairments in people with diabetes. Stroke has been associated with greater income losses than other health conditions.7
There are several other NCDs that have not been as frequently studied for their association with disability, for example, dementia, gout, traumatic brain injury (TBI) and mental health (MH) conditions. Further research on disability could benefit from using disability measures that can be compared across different health conditions, rather than measures only relevant to a single disease.5 Few studies appear to examine the impact of comorbidity on disability, and social inequalities in disability measures are not routinely examined.
Linked health, census and tax data in Statistics NZ’s Integrated Data Infrastructure (IDI) provides the opportunity to examine the association between NCDs and disability in a national population, with subgroup stratification to examine for inequalities.8 There have been a range of different outcomes from studies in different contexts, with different measures of NCD prevalence and definitions of disability.5 This research can assist policy makers to characterise NCD associations in this setting using specified measures of disability. Consideration of NCD effects by ethnicity and deprivation may help to understand the impact of resources on a person’s ability to continue to function with a particular condition.
The aim of this study is to measure how common functional impairment is for individuals with NCDs; and whether the level of functional impairment differs by health condition, comorbidity, age, sex, ethnicity and deprivation. Frequency in patterns of overlap between each NCD is examined to explore the implications of comorbidities on disability.
Methods
Study design
This is a cross-sectional study of people living in NZ which uses data from the Statistics NZ IDI, a national research database of deterministically and probabilistically linked administrative, survey and other data sets.9 The IDI comprises a central spine representing the ever-residential population of NZ, created by probabilistic linkage of tax, birth or travel visa records.9 National health data sets were linked together deterministically using a national health index data. Then in each refresh health data, and census data and other data sets are linked probabilistically to the spine. Each of these data sets has different linkage rates and linkage quality for linkage to the spine. For the September 2021 refresh, for example, link rates to the IDI spine were 95% for the 2018 census and 85% for health; with 0.9% and 0.8% false positive rates respectively.10
Study population
Residents were identified from the IDI estimated residential population on 30 June 2018.11 This identified 4.79 million residents, which is only slightly less than the official Statistics NZ resident population estimate of 4.90 million. This involves selecting individuals of any age from the IDI spine if they have had any activity in health, tax or education data sets in the year prior. Individuals were excluded from the analysis if they migrated or died before this date. The size of this population is within 2% of official population estimates (as seen previously)11 and has a better coverage than the 2018 census (4.18 million census records were linked to the IDI),12 particularly for minority groups.
Sociodemographic characteristics
Sociodemographic information, specifically age (0–14, 15–34, 35–54, 55–64, 65–74, 75+ years on 30 June 2018), sex (male, female) and ethnicity at ‘level 1’ of the NZ Standard Classification of Ethnicity (Asian, European, Māori, Middle Eastern, Latin American and African (MELAA), Pacific and Other), were identified from the personal details table (summary IDI table drawing on cross agency data). Ethnicity at ‘level 2 for groups comprising more than 5% of the population (ie, Chinese and Indian) was identified where available from the 2018 census, 2013 census and Admin Population Census (summary IDI table drawing on cross agency data) in that order of priority. People with two or more ethnicities contributed to all the ethnicity groups to which they identified in the ethnicity-specific analyses (total ethnicity approach). NZDep18,13 a deprivation index based on area of residence in the 2018 census, was assigned in quintiles to each individual using their most recent address notification up to 30 June 2018. Living situation was assigned using the 2018 census data according to whether an individual had information available about their family and household composition.
Non-communicable diseases
National health data sets were searched to identify nine NCDs, from 5 March 2008 up to 6 March 2018 (the date of the 2018 census). Data sets comprised public and private hospital discharges (NMDS), specialty mental health service contacts (PRIMHD), disability needs assessment and service coordination information (SOCRATES), needs assessment for older people (interRAI), pharmaceutical dispensing, laboratory claims, cancer registry, outpatient visits (NNPAC), Accident Compensation Corporation data set of accident-injury claims (ACC) and the chronic conditions table (IDI summary table) (online supplemental table A).
Nine NCDs were selected based on their inclusion in the NZ Ministry of Health classification of chronic conditions (diabetes, cancer, chronic obstructive pulmonary disease (COPD), coronary heart disease (CHD, including acute myocardial infarction), gout, stroke and TBI with the addition of dementia14 and MH/behavioural conditions.15 16 The latter was a broad group including attention deficit hyperactivity disorder, anxiety, autism spectrum disorder, bipolar disorder, conduct disorder, depression, eating disorders, emotional problems, personality disorders, psychotic disorders and sleep disorders (but not drug or alcohol disorders, or self-harm). Further information on NCD definitions is available in online supplemental table B.8
Functional impairment
Census, hospital discharges and tax data were used to select four indices of functional impairment. These measures were frequency or cumulative length of hospital stays (in the year prior), multiple comorbidities identified from hospitalisation data using the Elixhauser comorbidity index (in the 5 years prior), self-reported activity limitation (2018 census, with analysis limited to the 73.2% of the population with census records) and income support (income-tested benefit, in working age population, in the year prior to 30 June 2018). See online supplemental table C,D and online supplemental figure A for precise definitions. These indicators were selected because they were relevant across all health conditions, and they covered a range of disability impacts including functional impairment, economic effects and need for health services. If there was no linked hospitalisation or benefit data, then we assumed that these indicators were not present.
Data management
Population, demographic, NCD and functional impairment data sets were combined using the IDI individual unique identifier. This identifier is assigned by Statistics NZ using deterministic linkage within agency data sets (eg, using the NHI to link the various health data sets) and by probabilistic linkage between distinct agency data sets (eg, links between health and the spine, and between census and the spine).17
Analysis
Descriptive tables and figures were produced to examine overall and sociodemographic patterns in the prevalence of NCDs by age, sex, ethnicity and deprivation. Level 2 ethnic groups were selected to explore ethnic comparisons in greater depth. Age-specific NCD prevalence rates were examined by ethnicity and deprivation. The level of functional impairment was reported by type of NCD, and by sociodemographic factors. Age-specific levels of activity limitation were examined by NCD type. Logistic regression was used to examine the independent association of NCDs, sex, age, ethnicity and deprivation quintile with functional impairment outcomes—hospitalisation, activity limitation and benefit receipt. Crude and adjusted OR and 95% CIs are reported. Finally, the overlaps between types of NCDs were examined to further reflect on patterns of comorbidity (including stratified by ethnicity).
All analyses were carried out in a secure IDI environment (Datalab), using SAS Enterprise Guide V.8.3. Confidentiality rules in the IDI required suppression of small numbers (<6) and random rounding of all counts to base 3, which means some totals may not precisely add up. Missing data were presented in tables where numbers were large enough to do this.
Results
Non-communicable diseases
On 30 June 2018, we identified 4.79 million NZ residents, including 1.38 million (28.8%) who had one or more NCDs (table 1). TBI (0.47 m, 9.8%), MH and behavioural conditions (0.35 m, 7.3%) and diabetes (0.32 m, 6.8%) were the most identified health conditions. NCDs were more prevalent in men than women overall (31.4% vs 26.1%), increased with age and increased with deprivation (online supplemental table E). A high proportion of people living alone had a NCD (46.9%) and rates of all specific conditions were higher for those living alone (likely due to age), except for TBI. There was variation in NCD disparities by ethnicity and deprivation across the age groups. For example, in the young adult age group (15–34 years), Chinese and Indian ethnic groups had the lowest prevalence rates of NCDs, while Māori, European and Pacific peoples had the highest rates (figure 1). In the older age groups (55+ years), Chinese, European and Other ethnic groups tended to have the lowest rates, while Pacific, Indian and Māori had the highest prevalence rates. The greater prevalence of NCDs by deprivation was much more marked in adult age groups (especially in 35+ year olds) but was not evident in children (online supplemental figure B).
Table 1. Prevalence of non-communicable disease by sociodemographic distribution in Aotearoa New Zealand, 2018.
| Individuals | Distribution(column %) | Any condition (%) | 2 or more conditions(%) | ||
| All | All | 4 789 287 | 100.00 | 28.8 | 7.8 |
| Sex | Male | 2 384 676 | 49.8 | 31.4 | 8.8 |
| Female | 2 404 611 | 50.2 | 26.1 | 6.9 | |
| Age (years) | 0–14 | 917 445 | 19.2 | 14.5 | 0.9 |
| 15–34 | 1 333 584 | 27.8 | 23.2 | 3.5 | |
| 35–54 | 1 228 659 | 25.7 | 25.4 | 5.3 | |
| 55–64 | 570 819 | 11.9 | 36.6 | 10.7 | |
| 65–74 | 425 634 | 8.9 | 48.8 | 18.8 | |
| 75+ | 313 143 | 6.5 | 65.9 | 36.5 | |
| Ethnic group | European | 3 416 328 | 71.3 | 30.8 | 8.4 |
| Māori | 857 064 | 17.9 | 30.9 | 8.1 | |
| Pacific | 426 123 | 8.9 | 29.3 | 7.7 | |
| Asian | 751 590 | 15.7 | 17.4 | 3.6 | |
|
246 753 | 5.2 | 16.4 | 3.2 | |
|
251 913 | 5.3 | 19.9 | 4.6 | |
| MELAA | 96 894 | 2.0 | 22.3 | 5.2 | |
| Other ethnicity | 120 351 | 2.5 | 29.7 | 7.4 | |
| NZDep18 quintile | Lowest | 909 507 | 19.0 | 25.4 | 5.5 |
| Low-middle | 924 672 | 19.3 | 27.2 | 6.8 | |
| Middle | 946 758 | 19.8 | 28.3 | 7.6 | |
| High-middle | 974 343 | 20.3 | 30.3 | 8.9 | |
| Highest | 1 020 243 | 21.3 | 32.5 | 10.0 | |
| Missing | 13 761 | 0.3 | 2.9 | 0.6 | |
| Living situation* | One person household | 360 954 | 7.5 | 46.9 | 19.2 |
| Multiperson household | 316 269 | 6.6 | 29.6 | 8.1 | |
| Family/extended family | 3 362 883 | 70.2 | 26.9 | 6.3 | |
| Data not available | 749 178 | 15.6 | 27.8 | 9.1 |
Notes: MELAA, Middle Eastern, Latin American and African. Random rounding to base three according to confidentiality requirements may cause some totals to not perfectly add up. Distribution column percentages add up to more than 100% because some individuals will identify with more than one ethnic group. Note different age structures will influence the ethnicity comparisons in this table and have not been adjusted for in this descriptive table (Ssee figure 1). In addition, Supplementary online supplemental table HTable H reports prevalence of each health condition. Data were provided by Statistics New Zealand. *For those recorded by the Census. Data was provided by Statistics .
For those recorded by the 2018 Census.
MELAAMiddle Eastern, Latin American and African
Figure 1. Ethnic differences in non-communicable disease prevalence by age, in Aotearoa New Zealand, 2018. Data were provided by Statistics New Zealand. MELAA, Middle Eastern, Latin American and African; NCDs, non-communicable disease.
Among the group of people with an NCD for whom, there was living situation data, 12.0% lived alone, 65.8% lived with family and the remainder lived in another type of multiperson household. Young women were more likely to live with family than young men but from age 55 years, more men lived with family than women did. More Pacific peoples and Māori lived with family than other ethnic groups.
Functional impairment
In the total population, 1.7% had high levels of hospitalisation, 5.6% had a high multiple comorbidity score (Elixhauser 2+), 6.4% had an activity limitation (of available records) and 16.6% received income support (of working-age people) (table 2). There was a clear pattern of greater impairment in people with more NCDs and the pattern persisted after adjustment for age, sex, ethnicity and deprivation (table 3). People with two or more NCDs, compared with those with no NCDs, had 12 times the odds of high levels of hospitalisation (adjusted OR 12.0, CI: 11.8 to 12.3), 4 times the odds of an activity limitation (aOR 4.4, CI: 4.4 to 4.5) and nearly 5 times the odds of receiving income support (aOR 4.8, CI: 4.7 to 4.8).
Table 2. Level of functional impairment by non-communicable disease and sociodemographic factors, in Aotearoa New Zealand, 2018.
| Total population | High level of hospitalisation,*n (%) | Elixhauser score 2+,N (%) | Any activity limitation,‡n (%) | Any income support,§n (%) | ||
| Totals | All | 4 789 287 | 80 457 (1.7) | 268 266 (5.6) | 223 539 (6.4) | 478 674 (16.6) |
| Number of NCDs | 0 | 3 411 855 | 19 305 (0.6) | 33 930 (0.9) | 83 208 (3.4) | 250 569 (11.8) |
| 1 | 1 002 618 | 22 800 (2.3) | 89 790 (1.9) | 66 603 (8.6) | 156 177 (25.7) | |
| 2+ | 374 814 | 38 352 (10.2) | 144 546 (38.6) | 73 728 (25.1) | 71 928 (43.3) | |
| Any NCD | 1 377 432 | 61 152 (4.4) | 234 336 (17.0) | 140 331 (13.1) | 228 105 (29.5) | |
| Condition | Cancer | 182 055 | 17 658 (9.7) | 60 609 (33.3) | 26 649 (16.8) | 12 792 (20.1) |
| COPD | 194 589 | 14 838 (7.6) | 52 593 (27.0) | 33 357 (20.5) | 26 235 (28.2) | |
| CHD | 149 424 | 17 379 (11.6) | 76 782 (51.4) | 33 000 (26.1) | 12 153 (29.5) | |
| Dementia | 28 674 | 6045 (21.1) | 12 504 (43.6) | 15 198 (67.8) | 1014 (61.8) | |
| Diabetes | 324 450 | 19 416 (6.0) | 103 110 (31.8) | 45 207 (17.1) | 47 595 (27.3) | |
| Gout | 172 275 | 10 836 (6.3) | 44 682 (25.9) | 23 676 (17.2) | 21 132 (23.9) | |
| Stroke | 42 429 | 7656 (18.0) | 27 114 (63.9) | 14 556 (42.3) | 4914 (44.8) | |
| TBI | 470 163 | 15 303 (3.3) | 43 044 (9.2) | 33 015 (9.7) | 77 121 (29.4) | |
| MH | 349 350 | 25 644 (7.3) | 66 069 (18.9) | 47 211 (19.3) | 117 846 (49.0) | |
| Sex | Male | 2 384 679 | 37 395 (1.6) | 134 349 (5.6) | 106 983 (6.3) | 226 953 (15.7) |
| Female | 2 404 611 | 43 062 (1.8) | 133 917 (5.6) | 116 553 (6.4) | 251 721 (17.4) | |
| Age (years) | 0–14 | 917 448 | 7386 (0.8) | 6273 (0.7) | 13 932 (2.9) | – |
| 15–34 | 1 333 590 | 12 285 (0.9) | 23 604 (1.8) | 31 284 (3.3) | 196 548 (18.0) | |
| 35–54 | 1 228 659 | 12 456 (1.0) | 44 355 (3.6) | 39 108 (4.0) | 185 442 (15.1) | |
| 55–64 | 570 819 | 9492 (1.7) | 44 793 (7.8) | 34 080 (7.1) | 96 687 (16.9) | |
| 65–74 | 425 634 | 12 795 (3.0) | 60 006 (14.1) | 37 509 (10.3) | – | |
| 75+ | 313 146 | 26 040 (8.3) | 89 235 (28.5) | 67 623 (25.9) | – | |
| Ethnicity | Māori | 857 064 | 14 874 (1.7) | 48 477 (5.7) | 39 015 (7.8) | 171 414 (35.9) |
| Pacific | 426 123 | 6660 (1.6) | 23 703 (5.6) | 17 010 (7.5) | 58 485 (25.1) | |
| Asian | 751 587 | 5739 (0.8) | 20 340 (2.7) | 17 907 (3.5) | 37 620 (7.4) | |
| Chinese | 246 753 | 1653 (0.7) | 5727 (2.3) | 6351 (3.5) | 9156 (5.7) | |
| Indian | 251 913 | 2379 (0.9) | 9084 (3.6) | 6174 (3.8) | 12 438 (6.9) | |
| MELAA | 96 894 | 1182 (1.2) | 3465 (3.6) | 2406 (4.4) | 14 091 (21.2) | |
| Other ethnicity | 120 348 | 1719 (1.4) | 6207 (5.2) | 6153 (6.6) | 12 483 (15.5) | |
| European | 3 416 325 | 62 985 (1.8) | 202 890 (5.9) | 174 732 (6.5) | 311 319 (15.7) | |
| NZ Deprivation Index 2018 quintile | Lowest | 909 507 | 10 470 (1.2) | 35 865 (3.9) | 27 639 (3.7) | 36 204 (6.7) |
| Low-middle | 924 678 | 13 596 (1.5) | 44 706 (4.8) | 36 960 (5.1) | 55 185 (9.9) | |
| Middle | 946 758 | 15 717 (1.7) | 51 543 (5.4) | 43 818 (6.1) | 76 830 (13.3) | |
| High-middle | 974 343 | 18 747 (1.9) | 61 974 (6.4) | 53 811 (7.7) | 112 626 (18.9) | |
| Highest | 1 020 246 | 21 900 (2.1) | 74 115 (7.3) | 61 272 (9.8) | 197 643 (32.8) | |
| Missing | 13 764 | 27 (0.2) | 69 (0.5) | 39 (7.1) | 189 (2.0) | |
| Living situation† | Living in one-person household | 360 954 | 15 117 (4.2) | 53 532 (14.8) | 46 107 (13.7) | 49 668 (26.6) |
| Living in multiperson household | 316 269 | 4974 (1.6) | 17 667 (5.6) | 15 750 (6.5) | 52 197 (19.7) | |
| Living in a family/extended family | 3 362 880 | 42 714 (1.3) | 152 205 (4.5) | 144 492 (5.0) | 249 345 (12.8) | |
| Living situation data not available | 749 184 | 17 649 (2.4) | 44 859 (6.0) | 17 187 (38.4) | 127 464 (26.4) |
Online supplemental table 1 records the proportion of the population in each stratum with available data for these measures. Different age structures will affect the comparisons in this table, which have not been age-standardised. Data were provided by Statistics New Zealand.
4+ visits/stays or 10+ cumulative days in 2017/2018.
For those recorded by the 2018 census.
Denominator is the 2018 census records, aged over 5 years.
Data were available in 2017/2018 for 18–64 year olds, and this age group is the denominator for the income support indicator.
CHD, coronary heart disease; COPD, chronic obstructive pulmonary disease; MELAAMiddle-Eastern Latin-American or AfricanMH, mental healthNCDnon-communicable diseaseTBItraumatic brain injury
Table 3. Relative odds of functional impairment by non-communicable disease and sociodemographic status, Aotearoa New Zealand, 2018.
| High level of hospitalisation* | Any activity limitation† | Any income support | |||||
| Crude | Adjusted | Crude | Adjusted | Crude | Adjusted | ||
| Number of NCDs(adjusted for age, sex, ethnicity, deprivation) | 0 | 1 | 1 | 1 | 1 | 1 | 1 |
| 1 | 4.09 (4.01 to 4.17) | 3.44 (3.37 to 3.51) | 2.65 (2.62 to 2.68) | 2.00 (1.98 to 2.02) | 2.58 (2.56 to 2.60) | 2.30 (2.29 to 2.32) | |
| 2+ | 20.03 (19.68 to 20.39) | 12.02 (11.77 to 12.27) | 9.50 (9.40 to 9.60) | 4.42 (4.36 to 4.47) | 5.68 (5.62 to 5.74) | 4.75 (4.69 to 4.80) | |
| N | 4 789 287 | 4 720 740 | 3 503 691 | 3 498 249 | 2 889 174 | 2 842 575 | |
| Sex(adjusted for age) | Male | 1 | 1 | 1 | 1 | 1 | 1 |
| Female | 1.15 (1.13 to 1.16) | 1.06 (1.04 to 1.07) | 1.02 (1.01 to 1.02) | 0.97 (0.96 to 0.98) | 1.13 (1.13 to 1.14) | 1.14 (1.13 to 1.15) | |
| N | 4 789 287 | 4 789 287 | 3 503 691 | 3 503 691 | 2 889 174 | 2 889 174 | |
| Age (years)(adjusted for sex) | 0–14 | 0.79 (0.77 to 0.82) | 0.79 (0.77 to 0.82) | 0.73 (0.72 to 0.75) | 0.73 (0.72 to 0.75) | NA | NA |
| 15–34 | 0.91 (0.89 to 0.93) | 0.91 (0.89 to 0.93) | 0.84 (0.82 to 0.85) | 0.83 (0.82 to 0.85) | 1.24 (1.23 to 1.25) | 1.24 (1.23 to 1.25) | |
| 35–54 | 1 | 1 | 1 | 1 | 1 | 1 | |
| 55–64 | 1.65 (1.61 to 1.70) | 1.65 (1.61 to 1.70) | 1.87 (1.84 to 1.90) | 1.87 (1.84 to 1.90) | 1.15 (1.14 to 1.16) | 1.15 (1.14 to 1.16) | |
| 65–74 | 3.03 (2.95 to 3.10) | 3.03 (2.95 to 3.10) | 2.80 (2.76 to 2.84) | 2.80 (2.76 to 2.84) | NA | NA | |
| 75+ | 8.86 (8.67 to 9.05) | 8.83 (8.64 to 9.02) | 8.49 (8.38 to 8.61) | 8.50 (8.39 to 8.62) | NA | NA | |
| N | 4 789 287 | 4 789 287 | 3 503 691 | 3 503 691 | 2 889 174 | 2 889 174 | |
| Total response ethnicity(adjusted for age, sex, deprivation) | Māori | 0.87 (0.86 to 0.89) | 1.31 (1.28 to 1.33) | 1.05 (1.03 to 1.06) | 1.50 (1.48 to 1.52) | 3.07 (3.05 to 3.09) | 2.41 (2.39 to 2.43) |
| Pacific | 0.78 (0.76 to 0.80) | 1.15 (1.12 to 1.19) | 0.89 (0.87 to 0.91) | 1.29 (1.27 to 1.32) | 1.44 (1.42 to 1.46) | 1.10 (1.09 to 1.11) | |
| Asian | |||||||
|
0.30 (0.29 to 0.32) | 0.50 (0.47 to 0.53) | 0.35 (0.34 to 0.36) | 0.64 (0.62 to 0.66) | 0.27 (0.27 to 0.28) | 0.35 (0.34 to 0.36) | |
|
0.43 (0.41 to 0.45) | 0.76 (0.73 to 0.80) | 0.38 (0.37 to 0.39) | 0.71 (0.69 to 0.73) | 0.33 (0.32 to 0.34) | 0.33 (0.32 to 0.34) | |
|
0.31 (0.30 to 0.33) | 0.59 (0.56 to 0.62) | 0.34 (0.33 to 0.35) | 0.69 (0.67 to 0.71) | 0.46 (0.45 to 0.47) | 0.50 (0.49 to 0.51) | |
| MELAA | 0.61 (0.57 to 0.65) | 1.05 (0.99 to 1.11) | 0.49 (0.47 to 0.51) | 0.98 (0.93 to 1.02) | 1.29 (1.26 to 1.31) | 1.39 (1.36 to 1.41) | |
| Other ethnicity | 0.79 (0.76 to 0.83) | 0.94 (0.90 to 0.99) | 0.96 (0.94 to 0.99) | 1.18 (1.15 to 1.22) | 0.97 (0.95 to 0.99) | 1.02 (1.00 to 1.05) | |
| European | 0.86 (0.84 to 0.88) | 1.00 (0.97 to 1.02) | 0.66 (0.65 to 0.68) | 0.87 (0.86 to 0.88) | 0.69 (0.69 to 0.70) | 0.91 (0.90 to 0.92) | |
| N | 4 730 052 | 4 720 740 | 3 498 795 | 3 498 249 | 2 848 221 | 2 842 575 | |
| NZ Deprivation Index 2018 quintile(adjusted for age, sex, ethnicity) | Lowest | 1 | 1 | 1 | 1 | 1 | 1 |
| Low-middle | 1.28 (1.25 to 1.32) | 1.23 (1.20 to 1.26) | 1.38 (1.36 to 1.41) | 1.32 (1.30 to 1.34) | 1.53 (1.51 to 1.56) | 1.52 (1.50 to 1.54) | |
| Middle | 1.45 (1.41 to 1.49) | 1.39 (1.35 to 1.42) | 1.69 (1.66 to 1.71) | 1.60 (1.57 to 1.62) | 2.14 (2.12 to 2.17) | 2.09 (2.06 to 2.12) | |
| High-middle | 1.68 (1.64 to 1.73) | 1.59 (1.56 to 1.63) | 2.16 (2.13 to 2.19) | 2.00 (1.97 to 2.04) | 3.27 (3.23 to 3.31) | 3.03 (2.99 to 3.07) | |
| Highest | 1.88 (1.84 to 1.93) | 1.91 (1.86 to 1.96) | 2.81 (2.77 to 2.85) | 2.63 (2.59 to 2.67) | 6.85 (6.77 to 6.93) | 5.31 (5.24 to 5.38) | |
| N | 4 775 523 | 4 720 740 | 3 503 142 | 3 498 249 | 2 879 766 | 2 842 575 | |
The crude analysis is a logistic regression model output for just one explanatory variable, and the adjusted model includes the additional variables reported for each variable in column one. Total response ethnicity ORs compare individuals who identify with that ethnicity with those who do not identify with that ethnicity. When ethnicity was in the crude and multivariable models, all eight ethnicity groupings were included. Chinese, Indian and Other Asian are subsets of Asian. Odds ratios are presented here with their 95% confidence intervals. Data were provided by Statistics New Zealand.
4+ hospital visits/stays or 10+ cumulative days in 2017/2018.
2018 census question.
MELAAMiddle-Eastern Latin-American or AfricanNtotal number of individuals with data input into in regression model aboveNCDnon-communicable disease
Levels of functional impairment varied substantially by age, ethnicity and deprivation. Activity limitation for example was similar by sex, but much more common in older people (25.9% in 75+ year olds vs 2.9% in 5–14 year olds); and more common with deprivation (9.8% in the most deprived vs 3.7% in the least deprived quintiles). Māori (aOR 1.50, 95% CI: 1.48 to 1.52) and Pacific populations (aOR 1.29, CI: 1.26 to 1.31) experienced the highest levels of activity limitation compared with non-Māori and non-Pacific, respectively, with the lowest activity limitation levels in Asian compared with non-Asian (aOR 0.68, CI: 0.66 to 0.70). When the analysis was confined to individuals with an NCD, levels of self-reported activity limitation were more similar across ethnicity groups (eg, Māori 14.2%, Pacific 12.9%, Asian 8.6%, European 13.2%). People living alone also had high rates of activity limitation (13.7% vs 5.0% for people living in a family).
The levels of functional impairment differed by NCD type. Activity limitation was more common in people with dementia, stroke and CHD across all measures (except for CHD with income support receipt). For these conditions, there were high levels of hospitalisation (4+ visits/stays or 10+ cumulative days in 2017/2018; 21%, 18% and 12%, respectively), a two or more Elixhauser index (44%, 64% and 51%, respectively), and activity limitation (68%, 42% and 26%, respectively). Activity limitation was also very high for MH conditions, and indeed higher than CHD, across all age groups when stratified by age (figure 2). Income support was high in people with dementia and stroke (62% and 45%), and in people with a MH condition (49%). Activity limitation measures were lowest in people with cancer (then TBI and gout) after stratification for age (figure 2). Working-age people with cancer and gout had the lowest levels of income support (20% and 24%).
Figure 2. Activity limitation by non-communicable disease and age group, Aotearoa New Zealand, 2018. No condition means that none of the other conditions listed here were identified. Data were provided by Statistics New Zealand, from the 2018 Census. CHD, coronary heart disease; COPD, chronic obstructive pulmonary disease; MH, mental health; NCD, non-communicable disease; TBI, traumatic brain injury.
Walking and climbing steps were the most likely activity limitation in people with NCDs (15% in the group with 2+ NCDs) (online supplemental table F). For people with dementia, self-reported remembering or concentrating difficulties were common (54%), as well as difficulties with washing (39%) and difficulties with walking or climbing steps (39%). Online supplemental table F shows a mobility predominant activity limitation pattern for most NCDs, whereas Dementia, TBI and MH conditions had more of a remembering/concentrating predominant limitation pattern.
Comorbidities
The overlap of study specified NCDs is shown in table 4. Individuals with dementia, stroke and CHD were more likely to have at least one other NCD (87%, 78% and 71%, respectively). In people with dementia, the most common overlapping condition was MH and behavioural conditions (affecting 51%) followed by CHD (32%). In people who have had a stroke, 30% have also had diabetes and 30% have had CHD. In people with CHD, 32% also have had diabetes. The rates of diabetes overlapping with stroke and CHD were much greater when stratified by ethnicity, for example, in Indian (60% and 66%), Pacific (55% and 52%), Chinese (43% and 47%) and Māori (39% and 44%) compared with European (26% and 27%) (online supplemental table G). Other patterns of interest were the high rate of TBI among people with dementia (26%) and MH conditions (21%).
Table 4. Proportion of people with a non-communicable disease, who have another non-communicable disease, all ages, Aotearoa New Zealand, 2018.
| Subgroup | Proportion of each subgroup who also have the health condition below (%) | 2+ health conditions (%) | Mean number of health conditions | Total | ||||||||
| Dementia | Stroke | CHD | Gout | Cancer | COPD | Diabetes | MH | TBI | ||||
| Dementia | 18.7 | 32.3 | 12.5 | 22.2 | 22.4 | 24.3 | 50.7 | 25.6 | 87.4 | 3.09 | 28 671 | |
| Stroke | 12.6 | 29.7 | 16.6 | 19.3 | 20.2 | 30.3 | 26.9 | 20.2 | 78.2 | 2.76 | 42 429 | |
| CHD | 6.2 | 8.4 | 18.1 | 17.4 | 21.5 | 32.1 | 15.6 | 13.0 | 70.5 | 2.32 | 149 424 | |
| Gout | 2.1 | 4.1 | 15.7 | 12.9 | 13.5 | 30.9 | 9.5 | 10.0 | 59.3 | 1.99 | 172 269 | |
| Cancer | 3.5 | 4.5 | 14.2 | 12.2 | 13.8 | 18.2 | 10.5 | 10.2 | 51.3 | 1.87 | 182 052 | |
| COPD | 3.3 | 4.4 | 16.5 | 11.9 | 12.9 | 21.0 | 14.3 | 11.5 | 55.0 | 1.96 | 194 592 | |
| Diabetes | 2.2 | 4.0 | 14.8 | 16.4 | 10.2 | 12.6 | 11.7 | 9.3 | 50.8 | 1.81 | 324 447 | |
| MH | 4.2 | 3.3 | 6.7 | 4.7 | 5.5 | 7.9 | 10.8 | 21.3 | 41.5 | 1.64 | 349 350 | |
| TBI | 1.6 | 1.8 | 4.1 | 3.7 | 4.0 | 4.8 | 6.4 | 15.8 | 28.6 | 1.42 | 470 169 | |
| Total population | 0.6 | 0.9 | 3.1 | 3.6 | 3.8 | 4.1 | 6.8 | 7.3 | 9.8 | 7.8 | 1.40 | 4 789 287 |
Note: COPD, ; CHD, coronary disease; MH, mental health, and behavioural conditions; TBI, .Ordered from least to most common condition. Patterns will be affected by the mean age of participants in each health condition group. Data wasere provided by Statistics New Zealand.
CHDcoronary heart diseaseCOPDchronic obstructive pulmonary diseaseMHmental healthTBItraumatic brain injury
Discussion
Summary of results
Almost three-tenths of the study population (29%) of nearly 5 million people experienced at least one of the nine NCDs during the study period (2008–2018), and prevalence rates were greater in men, Māori and Pacific, and people living in deprived areas. Prevalence increased to two-thirds (66%) in the group who were 75 or more years old. One mediator of differences in NCD prevalence is the distribution of key risk factors such as tobacco use, diet and obesity, and alcohol use. In the 2017/2018 Health Survey, for example, 15% of adults were current smokers, 32% of adults were obese and 20% of adults drank alcohol in a way that could harm themselves of others. All these risk factors were more common in men than women, in Māori and Pacific than European, and more common in adults living in areas with the highest deprivation.
Functional impairment was 2–3 times more likely in individuals with a NCD, and 4–12 times more likely if two or more NCDs were present—independent of age, sex, ethnicity and deprivation. People with dementia, stroke and CHD were found to have the highest burden of comorbidity (eg, 87%, 79% and 71% had 2+ NCDs, respectively) and functional impairment (eg, prevalence of activity limitations was 68%, 42% and 26%, respectively). Stroke and CHD functional impairment findings are consistent with findings from a systematic review focused on high income countries,5 which indicates that people with CVD (including stroke and CHD) have a higher prevalence of disability in activities of daily living than people with cancer, chronic respiratory disease and diabetes. Our figures for activity limitation prevalence (cancer 17%, CHD 26%, stroke 42%, COPD 21% and diabetes 17%) were similar but often at the lower end of the range compared with that in the systematic review.5 The review reported high proportions of people with cancer (10%–35%), cardiovascular disease (21%–64%), chronic respiratory disease (7%–50%) and diabetes (12%–55%) who experienced difficulties with activities of daily living (ie, different proportions for eating, bathing and dressing for each NCD).
In addition to these conditions in the review, we also investigated NCDs: dementia, gout, TBI and MH.5 Our results highlight the severe impact of dementia and MH on functional impairment, greater than CHD and many other conditions after accounting for age (eg, figure 2). These conditions have a remembering/concentrating predominant limitation pattern and were associated with high levels of income support in the working age group (62% and 49%). Conversely, cancer (followed by TBI and gout) was associated with the lowest levels of activity limitation after accounting for age, but impairment levels were still double or more than that for people with no identified NCDs.
A previous NZ Treasury study examined the longitudinal impact of diagnosis with one of eight health conditions on employment, income support and income losses in the four years following diagnosis using propensity score modelling.7 The greatest impacts were found for stroke then TBI, CHD, diabetes, COPD and breast cancer in that order, providing strong evidence for a causal impact. The relatively greater impacts of stroke and CHD are consistent with our results, although we reported lower functional impacts from TBI. This may be because the effects of TBI are acute and then decrease over time,7 and our study of prevalence includes historical cases of TBI, for example, recorded up to 10 years prior, but possibly occurring even before this.
Multimorbidity was common in older age groups, particularly for conditions associated with the highest levels of functional impairment (dementia, stroke and CHD). The reported patterns of comorbid conditions are useful for considering additional healthcare needs for people with specific health conditions. The most common NCDs occurring together were MH conditions with dementia; and diabetes with stroke or CHD. There also may be common underlying risk factors that increase the risk of both conditions, such as age.
Strengths and limitations
The use of large national linked data sets improves the power of this study, enabling smaller stratifications (eg, by ethnicity and NCD), and providing a set of useful variables available at the national level to improve our understanding of functional impairment.
The cross-sectional nature of this study means that we cannot directly attribute disability to the underlying NCDs, or any demographic factor. However, we used multivariable analyses to adjust for the possibility that common underlying factors are driving the incidence of both NCDs and disability, such as age, sex, ethnicity and socioeconomic position.
Identification of NCDs was reliant on available health data sets, existing validated disease definitions, the validity of newly developed algorithms and the extent to which we can look-back at health records. We retrieved records going back 10 years (and 5 years for Needs assessment for older people), but the onset of some NCDs may have occurred before this, and been recorded later, for example, if there was ongoing treatment. Also, if someone does not access healthcare for a specific condition, our methods are unlikely to capture the corresponding NCD. Conversely, if someone is a frequent visitor to healthcare services, incidental diagnoses may be more likely to be captured. We were not able to include primary healthcare data, which may mean we have missed some mild cases not registered in any of the available national health data sets.
The selected measures of functional impairment and their coverage were limited to what was available from administrative and survey data sets. There was differential missing data for the 2018 census, which may have underestimated the levels of functional impairment in underrepresented groups such as Māori, Pacific and high deprivation, thus making our results here an underestimate of ethnic disparities.
Implications
The high prevalence of NCDs and their close link with functional impairment and disability (in this study and others) suggests that there is high value in NCD prevention for reducing the burden of disability. This is especially important for older persons, Māori and Pacific peoples, people living in deprived areas and people living alone; all of whom have more NCDs and higher levels of functional impairment.
This study characterises a large potential need for disability support, and further research could examine the level of provision and unmet need in this area. Most people with a NCD lived with family, posing a question about the impact of these NCDs on family members, who may be providing care and support that moderates functional impairment in this group. Future longitudinal studies would be useful to understand the precise relationship between chronic disease progression and functional impairment, identifying which factors lead to disability and any protective factors for disability (such as younger age, low deprivation and ethnicity). We recommend further standardisation and validation of NCD algorithm definitions in this setting.
Conclusions
NCDs are commonly experienced in this high-income country especially by older people. Functional impairment is strongly patterned by the prevalence of NCDs. Dementia, stroke, CHD and MH conditions are associated with the highest levels of functional impairment. Findings are likely relevant to high-income settings globally.
supplementary material
Acknowledgements
Thank you to those who shared code with the project team: Stephanie D’Souza (initial SAS code for chronic NCDs and ICD codes for dementia); Social Wellbeing Agency (SAS code for mental health and behavioural conditions). We thank the Public Policy Institute at The University of Auckland, the University of Otago Wellington and Statistics New Zealand Wellington Office for access to their Stats NZ Integrated Data Infrastructure data laboratories.These results are not official statistics. They have been created for research purposes from the Integrated Data Infrastructure (IDI) which is carefully managed by Stats NZ. For more information about the IDI please visit https://www.stats.govt.nz/integrated-data/. Data were provided by Statistics New Zealand, IDI project MAA2020-80.
Footnotes
Funding: This work was supported by Ministry of Business Innovation and Employment (MBIE), New Zealand grant number via three health National Science Challenges: Ageing Well, Better Start and Healthier Lives (Contract reference UOAX1901).
Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2023-079412).
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Not applicable.
Ethics approval: This study involves human participants. Ethics approval was given by the University of Auckland Health Research Ethics Committee (AH21563). We used a secure deidentified research database to do secondary data analysis. Consent is not possible and is not legally required for research done using this database, which is highly secure and all output is checked to ensure there is no identifying content.
Data availability free text: All data are available on application to the Statistics New Zealand Integrated Data Infrastructure, and are only accessible from a secure approved Data Lab environment.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting or dissemination plans of this research.
Contributor Information
Andrea M Teng, Email: andrea.teng@otago.ac.nz.
Lisa Underwood, Email: l.underwood@auckland.ac.nz.
Nicholas Bowden, Email: nick.bowden@otago.ac.nz.
Hamish Jamieson, Email: Hamish.Jamieson@otago.ac.nz.
Barry Milne, Email: b.milne@auckland.ac.nz.
Data availability statement
Data may be obtained from a third party and are not publicly available.
References
- 1.Vos T, Lim SS, Abbafati C. Global burden of 369 diseases and injuries in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet. 2020;396:1204–22. doi: 10.1016/S0140-6736(20)30925-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Disney G, Teng A, Atkinson J, et al. Changing ethnic inequalities in mortality in New Zealand over 30 years: linked cohort studies with 68.9 million person-years of follow-up. Popul Health Metr. 2017;15:1–15. doi: 10.1186/s12963-017-0132-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Blakely T, Kvizhinadze G, Atkinson J, et al. Health system costs for individual and comorbid noncommunicable diseases: An analysis of publicly funded health events from New Zealand. PLoS Med. 2019;16:e1002716. doi: 10.1371/journal.pmed.1002716. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Sarfati D, Koczwara B, Jackson C. The impact of comorbidity on cancer and its treatment. CA Cancer J Clin. 2016;66:337–50. doi: 10.3322/caac.21342. [DOI] [PubMed] [Google Scholar]
- 5.Lisy K, Campbell JM, Tufanaru C, et al. The prevalence of disability among people with cancer, cardiovascular disease, chronic respiratory disease and/or diabetes: a systematic review. Int J Evid Based Healthc. 2018;16:154–66. doi: 10.1097/XEB.0000000000000138. [DOI] [PubMed] [Google Scholar]
- 6.Üstūn B, Kennedy C. FORUM: THE ROLE OF FUNCTIONAL IMPAIRMENT IN THE DIAGNOSIS OF MENTAL DISORDERS: TOWARDS ICD-11 AND DSM-V: What is 'functional impairment'? Disentangling disability from clinical significance. World Psychiatry. 2009;8:82–5. doi: 10.1002/j.2051-5545.2009.tb00219.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Dixon S. New Zealand Treasury; 2015. The employment and income effects of eight chronic and acute health conditions. [Google Scholar]
- 8.Underwood L, Teng A, Bowden N, et al. Long-term health conditions among household families in Aotearoa New Zealand: cross-sectional analysis of integrated Census and administrative data. N Z Med J. 2024;137:20–34. doi: 10.26635/6965.6370. [DOI] [PubMed] [Google Scholar]
- 9.Milne BJ, Atkinson J, Blakely T, et al. Data Resource Profile: The New Zealand Integrated Data Infrastructure (IDI) Int J Epidemiol. 2019;48:677–677e. doi: 10.1093/ije/dyz014. [DOI] [PubMed] [Google Scholar]
- 10.Statistics New Zealand Statistical. Methods, September 2021 refresh. Integrated data infrastructure (IDI) refresh: linking report. 2021:1–12.
- 11.Gibb S, Bycroft C, Matheson-Dunning N. Identifying the New Zealand resident population in the integrated data infrastructure (IDI) New Zealand: Census Transformation Statistics; 2016. www.stats.govt.nz Available. [Google Scholar]
- 12.Statistics New Zealand Linking 2018 census respondents to the integrated data infrastructure. 2019. www.stats.govt.nz Available.
- 13.Atkinson J, Salmond C, Crampton P. Wellington: University of Otago; NZDep2018 index of deprivation: final research report, December 2020. [Google Scholar]
- 14.Walesby KE, Exeter DJ, Gibb S, et al. Prevalence and geographical variation of dementia in New Zealand from 2012 to 2015: Brief report utilising routinely collected data within the Integrated Data Infrastructure. Australas J Ageing. 2020;39:297–304. doi: 10.1111/ajag.12790. [DOI] [PubMed] [Google Scholar]
- 15.Richmond-Rakerd LS, D’Souza S, Milne BJ, et al. Longitudinal Associations of Mental Disorders With Physical Diseases and Mortality Among 2.3 Million New Zealand Citizens. JAMA Netw Open. 2021;4:e2033448. doi: 10.1001/jamanetworkopen.2020.33448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Bowden N, Gibb S, Thabrew H, et al. Case identification of mental health and related problems in children and young people using the New Zealand Integrated Data Infrastructure. BMC Med Inform Decis Mak. 2020;20 doi: 10.1186/s12911-020-1057-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Black A. The IDI prototype spines creation and coverage. Working paper no 16-03. Statistics New Zealand; www.stats.govt.nz Available. [Google Scholar]