Factors to consider when interpreting results
Estimates presented here may differ from those reported elsewhere due to differences in the data source, including differences in the method of data collection, as well as the specific chronic conditions included in analysis and how they are defined.
The conditions data used in this analysis are 'as reported' by respondents and may not necessarily represent conditions as medically diagnosed, or all conditions that a respondent may have. Similarly, estimates of the proportion of people who are sufficiently active are based on self-reported information which may affect interpretation.
Prevalence estimates based on self-reported conditions may differ from those reported based on diagnostic surveys or biomedical testing. In particular, estimates of people with self-reported mental or behavioural conditions presented from analysis of the NHS will differ from those obtained from a diagnostic tool such as that used in the 2007 National Survey of Mental Health and Wellbeing. The questions asked of NHS respondents are available in the National Health Survey questionnaire, 2017–18.
The degree to which reported conditions have been medically diagnosed is likely to differ across age groups and condition types. Where a condition has been reported, but is not medically diagnosed, overestimation of disease prevalence can occur.
The tendency to self-report a chronic condition can differ between individuals, based on characteristics of the individual such as age, cohort or language spoken at home, and may be influenced by characteristics of the condition such as whether it is episodic or persistent in nature (Dobson et al. 2020; Lujic et al. 2017).
In contrast, a person may have a condition and either not be aware of it, or be reluctant to report it due to personal preference or the personal or sensitive nature of the condition. Where a diagnosed condition is not self-reported, or multiple diagnosed conditions are selectively self-reported, disease prevalence may be underestimated. For example, it is known that self-reported data for chronic kidney disease substantially underestimate its prevalence compared with biomedical testing: 1.0% versus 10% respectively (AIHW 2014).
The implications of this for estimating multimorbidity, the co-prevalence of chronic conditions and the strength of the association between them are even more complex. As the tendency to self-report a chronic condition can differ between individuals, some people may be more likely to report a chronic condition than others, and if a person is more likely to report one condition, they may be more likely to report another (with or without a medical diagnosis). This may lead to the overestimation of multimorbidity, co-prevalence and of the strength of association between the conditions reported by the individual who is more likely to self-report a chronic condition.
Alternatively, where an individual may selectively report some diagnosed chronic conditions, but not others, or does not report any of the conditions they have been diagnosed with, multimorbidity, co-prevalence and the strength of the association between conditions may be underestimated.
The conditions included in this analysis were chosen for the substantial impact they have on people’s health, as well as data availability. As a result, this report can only estimate the prevalence of multimorbidity due to the 10 selected conditions and may underestimate the prevalence of multimorbidity that would be calculated if all chronic conditions could be included in the analysis.
When measuring the prevalence and impact of chronic conditions collectively, the inclusion (or exclusion) of specific conditions has the potential to affect results.
For example, hypertension is a highly prevalent, major risk factor for chronic conditions including stroke, heart failure and chronic kidney disease (AIHW 2019). If chronic hypertension was included in the analysis for selected cardiovascular diseases, this may have increased prevalence estimates for cardiovascular disease, and potentially estimates of multimorbidity prevalence. The inclusion of conditions more common in either females or males could influence findings about whether multimorbidity is more common in females or males.
The way that each chronic condition has been defined within this analysis may also influence estimates of multimorbidity. For example, cardiovascular disease has a large number of inclusions, such as stroke and heart attack, but a person who reports having had both a stroke and a heart attack are only counted as having 1 condition (selected cardiovascular diseases) when counting morbidities and are not counted as having multimorbidity. In contrast, osteoporosis is the only inclusion within that condition.
Research has shown that when multimorbidity is defined as having 2 or more ‘disease entities’, there is no substantial difference in prevalence estimates, regardless of whether a disease entity represents an individual chronic condition (such as osteoporosis) or has multiple inclusions (such as cardiovascular disease) when using International Classification of Primary Care (V.2; ICPC-2) chapter, International Classification of Disease (10th revision; ICD-10) chapter or Cumulative Illness Rating Scale (CIRS) domain (Harrison et al. 2014).
This suggests that the estimates of total multimorbidity presented here are likely to be less affected by the selected conditions and their inclusions than estimates of the prevalence of the number of chronic conditions experienced, and of complex multimorbidity types where classification is partly based on the number of conditions experienced.
How conditions are allocated to body systems will also influence estimates of complex multimorbidity types as the number of body systems affected defines, in part, whether someone will be categorised as having complex multimorbidity or not. ICD-10 chapters are used as a proxy for body system in this analysis (see Table 1 ‘Definitions used for chronic conditions’ in the section NHS chronic condition definitions for details). However, using a different classification, or a different method (such as expert advice) to allocate conditions to body systems could produce different results.
Chronic conditions are more common among older people. As a result, the likelihood of having multiple chronic conditions—multimorbidity—is also expected to be higher among older Australians. However, as the NHS samples from the non-institutionalised Australian population (ABS 2018), these results potentially underestimate the true prevalence and level of multimorbidity in the entire population, given that older people in institutionalised dwellings (such as residential care facilities) are not included.
It is important to note that while this analysis has found characteristics that may be more common in people with multimorbidity, it is not possible to definitively say that multimorbidity is caused by these characteristics with the current data. Similarly, it is not possible to say that the characteristics examined are the result of having multimorbidity. For example, while being overweight or obese may be a risk factor for multimorbidity, it could also be the result of a person being limited in the physical activity they can participate in as a result of their multimorbidity. Longitudinal information on when a person developed multimorbidity, and when they became overweight or obese would be needed to understand this, but is not available within the NHS.
ABS (Australian Bureau of Statistics) 2018. National Health Survey: First Results methodology, 2017–18. Canberra: ABS.
AIHW (Australian Institute of Health and Welfare) 2014. Cardiovascular disease, diabetes and chronic kidney disease—Australian facts: prevalence and incidence. Cardiovascular, diabetes and chronic kidney disease series no. 2. Cat. no. CDK 2. Canberra: AIHW.
AIHW 2019. High blood pressure. Cat. no. PHE 250. Canberra: AIHW. Viewed 23 February 2021.
Dobson A, Forder P, Hockey R, Egan N, Cavenagh D, Waller M et al. 2020. The impact of multiple chronic conditions: Findings from the Australian Longitudinal Study on Women’s Health. Report prepared for the Australian Government Department of Health, May 2020.
Harrison C, Britt H, Miller G & Henderson J. 2014. Examining different measures of multimorbidity, using a large prospective cross-sectional study in Australian general practice. BMJ Open 4(7): e004694.
Lujic S, Simpson JM, Zwar N, Hosseinzadeh H & Jorm L 2017. Multimorbidity in Australia: Comparing estimates derived using administrative data sources and survey data. PLoS ONE 12(8): e0183817.