National Health Survey: First Results methodology

This is not the latest release View the latest release
Reference period
2017-18 financial year
Released
12/12/2018

Explanatory notes

Introduction

1 This publication presents key indicators from the 2017-18 National Health Survey (NHS), including information on:

  • the health status of the population, including long-term health conditions;
  • health risk factors such as smoking, Body Mass Index, diet, exercise and alcohol consumption; and
  • demographic and socioeconomic characteristics.
     

2 The 2017-18 NHS was conducted throughout Australia from July 2017 to June 2018. Previous surveys were conducted in 1989-90, 1995, 2001, 2004-05, 2007-08, 2011-12 and 2014-15. Health surveys conducted by the ABS in 1977-78 and 1983, while not part of the NHS series, also collected similar information.

Scope of the survey

3 The NHS was conducted from a sample of approximately 21,300 people in 16,400 private dwellings across Australia.

4 Urban and rural areas in all states and territories were included, while Very Remote areas of Australia and discrete Aboriginal and Torres Strait Islander communities were excluded. These exclusions are unlikely to affect national estimates, and will only have a minor effect on aggregate estimates produced for individual states and territories, excepting the Northern Territory where the population living in Very Remote areas accounts for around 20.3% of persons.

5 Non-private dwellings such as hotels, motels, hospitals, nursing homes and short-stay caravan parks were excluded from the survey. This may affect estimates of the number of people with some long-term health conditions (for example, conditions which may require periods of hospitalisation or long term care).

6 Within each selected dwelling, one adult (18 years and over) and one child (0-17 years) were randomly selected for inclusion in the survey. This sub-sampling within households enabled more information to be collected from each respondent than would have been possible had all usual residents of selected dwellings been included in the survey. For the purposes of the NHS, a household was defined as one or more persons, at least one of whom is aged 18 years and over, usually resident in the same private dwelling.

7 The following groups were excluded from the survey:

  • certain diplomatic personnel of overseas governments, customarily excluded from the Census and estimated resident population;
  • persons whose usual place of residence was outside Australia;
  • members of non-Australian Defence forces (and their dependents) stationed in Australia; and
  • visitors to private dwellings.
     

Sample design

8 Dwellings were selected at random using a multistage area sample of private dwellings. The initial sample selected for the survey consisted of approximately 25,109 dwellings. This was reduced to a sample of 21,544 after sample loss (for example, households selected in the survey which had no residents in scope of the survey, vacant or derelict buildings, buildings under construction). Of those remaining dwellings, 16,384 (or 76.1%) were fully or adequately responding, yielding a total sample for the survey of 21,315 persons.

Approached sample, final sample and response rates

 NSWVic.QldSAWATas.NTACTAust.
Households approached
(after sample loss)
4 5373 4204 4121 9812 2231 7781 8281 36521 544
Households in sample3 2722 6143 3651 6591 6561 6061 0911 12116 384
Response rate (%)72.176.476.383.874.590.359.782.176.1
Persons in sample4 2733 4194 4122 0562 1682 0161 4791 49221 315

 

9 To take account of possible seasonal effects on health characteristics, the sample was spread across the 12-month enumeration period. Analysis of previous health surveys has shown no seasonal bias across key estimates.

Data collection

10 Trained ABS interviewers conducted personal interviews with selected residents in sampled dwellings. One adult (aged 18 years and over) in each dwelling was selected and interviewed about their own health characteristics as well as information about the household (for example, income of other household members). An adult, nominated by the household, was interviewed about one child in the household. Some children aged 15-17 years may have been personally interviewed with parental consent.

Weighting, benchmarking and estimation

11 Weighting is a process of adjusting results from a sample survey to infer results for the in-scope total population. To do this, a weight is allocated to each sample unit; for example, a household or a person. The weight is a value which indicates how many population units are represented by the sample unit. 

12 The first step in calculating weights for each person was to assign an initial weight, which was equal to the inverse of the probability of being selected in the survey. For example, if the probability of a person being selected in the survey was 1 in 600, then the person would have an initial weight of 600 (that is, they represent 600 others). An adjustment was then made to these initial weights to account for the time period in which a person was assigned to be enumerated.

13 The weights are calibrated to align with independent estimates of the population of interest, referred to as 'benchmarks', in designated categories of sex by age by area of usual residence. Weights calibrated against population benchmarks in this way compensate for over or under-enumeration of particular categories of persons and ensure that the survey estimates conform to the independently estimated distribution of the population by age, sex and area of usual residence, rather than to the distribution within the sample itself. 

14 The NHS was benchmarked to the estimated resident population living in private dwellings in non-Very Remote areas of Australia at 31 December 2017. Excluded from these benchmarks were persons living in discrete Aboriginal and Torres Strait Islander communities. The benchmarks, and hence the estimates from the survey, do not (and are not intended to) match estimates of the total Australian resident population (which include persons living in Very Remote areas or in non-private dwellings, such as hotels) obtained from other sources.

15 In 2017-18, data from the NHS and the Survey of Income and Housing (SIH) was combined to produce the National Health Survey and Survey of Income and Housing pooled dataset (NHS/SIH) and enable more accurate smoker status estimates. This dataset was also benchmarked to the above population to produce weights for this dataset. In addition, to preserve consistency between the two datasets, the NHS data was also benchmarked to the pooled NHS/SIH dataset by age, sex, area of usual residence and smoker status. This means that unperturbed smoker estimates will be identical between the NHS data and the NHS/SIH data at these cross-classifications.

16 Survey estimates of counts of persons are obtained by summing the weights of persons with the characteristic of interest. Estimates of non-person counts (for example, number of health conditions) are obtained by multiplying the characteristic of interest with the weight of the reporting person and aggregating.

Reliability of estimates

17 All sample surveys are subject to sampling and non-sampling error.

18 Sampling error is the difference between estimates, derived from a sample of persons, and the value that would have been produced if all persons in scope of the survey had been included. Indications of the level of sampling error are given by the Relative Standard Error (RSE) and 95% Margin of Error (MoE). For more information refer to the Technical Note - Reliability of Estimates.

19 In this publication, estimates with an RSE of 25% to 50% are preceded by an asterisk (e.g. *3.4) to indicate that the estimate has a high level of sampling error relative to the size of the estimate, and should be used with caution. Estimates with an RSE over 50% are indicated by a double asterisk (e.g. **0.6) and are generally considered too unreliable for most purposes.

20 Margins of Error are provided for proportions to assist users in assessing the reliability of these data. Estimates of proportions with an MoE more than 10% are annotated to indicate they are subject to high sample variability and particular consideration should be given to the MoE when using these estimates. Depending on how the estimate is to be used, an MoE greater than 10% may be considered too large to inform decisions. In addition, estimates with a corresponding standard 95% confidence interval that includes 0% or 100% are annotated with a # to indicate that they are usually considered unreliable for most purposes.

21 Non-sampling error may occur in any data collection, whether it is based on a sample or a full count such as a census. Non-sampling errors occur when survey processes work less effectively than intended. Sources of non-sampling error include non-response, errors in reporting by respondents or in recording of answers by interviewers, and errors in coding and processing data.

22 Non-response occurs when people are unable to or do not cooperate, or cannot be contacted. Non-response can affect the reliability of results and can introduce a bias. The magnitude of any bias depends on the rate of non-response and the extent of the difference between the characteristics of those people who responded to the survey and those who did not. 

23 In the 2017-18 NHS, measurements of height, weight and waist circumference were taken of respondents aged 2 years and over, while blood pressure was also measured for adult respondents (aged 18 years and over). While these items had relatively high non-response rates, analysis indicated no bias existed in the non-responding population. Imputation was used to obtain values for respondents for whom physical measurements were not taken. For more information see Appendix 2: Physical measurements in the 2017-18 National Health Survey.

24 The following methods were adopted to reduce the level and impact of non-response:

  • face-to-face interviews with respondents;
  • the use of proxy interviews in cases where language difficulties were encountered, noting the interpreter was typically a family member;
  • follow-up of respondents if there was initially no response; and
  • weighting to population benchmarks to reduce non-response bias;
     

Interpretation of results

25 Care has been taken to ensure that results are as accurate as possible. This includes thorough design and testing of the questionnaire, interviews being conducted by trained ABS Interviewers, and quality control procedures throughout data collection, processing and output. There remain, however, other factors which may have affected the reliability of results, and for which no specific adjustments can be made. The following factors should be considered when interpreting these estimates:

  • Information recorded in the survey is essentially 'as reported' by respondents, and hence may differ from information available from other sources or collected using different methodology; for example, information about health conditions is self-reported and, while not directly based on a diagnosis by a medical practitioner in the survey, respondents were asked whether they had ever been told by a doctor or nurse that they had a particular health condition. Conditions which have a greater effect on people's wellbeing or lifestyle, or those specifically mentioned in survey questions, are expected in general to have been better reported than others;
  • Some respondents may have provided responses that they felt were expected, rather than those that accurately reflected their own situation. Every effort has been made to minimise such bias through the development and use of appropriate survey methodology;
  • Results from previous surveys indicate a tendency for respondents to under-report consumption of alcohol; and
  • Under-reporting of young persons identifying as current smokers may have occurred due to social pressures, particularly in cases where other household members were present at the interview.
     

Comparability with previous National Health Surveys

26 Data for 2017-18 are comparable with earlier surveys, with some exceptions:

  • In 2017-18 an additional example "32. Learning difficulties, including dyslexia" was added to Prompt Card O2, for the Mental, Cognitive and Behavioural Conditions module. All other items remain the same and have been coded consistently with 2014-15 (as above);
  • In NHS 2017-18 a shorter version of the standard ABS Income questionnaire module for Household Surveys for the collection of 'Total Personal Income' and 'Total Household Income' was introduced. Overall data for 'Total Personal Income' and 'Total Household Income' is comparable between NHS 2017-18 and NHS 2014-15, however the breakdown by type of government pension is not available for NHS 2017-18 .
  • A new module regarding clients of the Department of Veterans' Affairs (DVA) has been added to NHS 2017-18, this should not be confused with the item used in previous NHS which presented data regarding persons who hold a "DVA Health Card";
  • Age ranges for two Disability items have been changed. In NHS 2014-15 the Item 'Whether has an education restriction' was limited to those aged 5 to 20 years, for 2017-18 this age range is persons aged 4 or more years, recognising that an education restriction can exist outside of school years and be life long. Similarly, the age range for 'Whether has an employment restriction' has been changed from 15 to 64 years to persons aged 15 years or more. Apart from this, the items are consistent with previous NHS;
  • New scales allowing measurements of up to 200kg were used in NHS 2017-18. In addition, a new stadiometer was used to measure height for greater accuracy. For this reason, an additional height measure was taken to analyse variation for Quality Assurance. Despite these changes, the estimates are considered comparable with 2014-15.
  • In line with Census 2016 a number of standard classifications used in the NHS have been updated in 2017-18, these include: Standard Australian Classification of Countries (SACC), 2016 (Country of Birth), Australian Standard Classification of Languages (ASCL) 2016 (Main language spoken at home), Australian and New Zealand Standard Industrial Classification (ANZSIC), 2006 - Coder 2018 (Industry of Main Job) and ANZSCO - Occupation Classification and Census of Population and Housing: Socio-Economic Indexes for Areas (SEIFA), Australia 2016. More information about these can be found on the ABS website. http://www.abs.gov.au/classifications
     

27 In 2014-15 and 2017-18, a module specifically dedicated to mental and behavioural conditions was included in the NHS to collect information on cognitive, organic and behavioural conditions. In previous NHS cycles, mental and behavioural conditions were collected in a module that included a wide range of long-term health conditions. The number of persons who reported having a mental and behavioural condition in 2014-15 increased from the 2011-12 NHS, potentially due to the greater prominence of mental and behavioural conditions in the new module. Data on mental and behavioural conditions for 2014-15 and 2017-18 are therefore not comparable with data in previous National Health Surveys.

28 Estimates of people with mental or behavioural conditions from 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.

29 For the 2017-18 NHS cycle, the smoking questionnaire module was used in both the NHS and the 2017-18 Survey of Income and Housing (SIH) to produce a larger sample size for more accurate smoker status estimates. The pooled dataset is known as the National Health Survey and Survey of Income and Housing (NHIH) and will contain data items common to both NHS and SIH such as age, sex, country of birth and those from the smoking module. In this publication, this pooled dataset is used whenever possible to produce estimates with smaller errors. The NHS dataset is used for items collected only in the NHS for example smoking status by BMI. The following table compares results produced from the NHIH and the NHS 2017-18 on its own. Note that the pooled dataset was used solely for smoker status and not consumption of cigarettes.

Smoking status, NHIH and NHS, 2017-18

National Health Survey and Survey of Income and Housing (pooled dataset)

 Estimate ('000)RSE of Estimate Proportion (%)MoE of Proportion 
Smoker status     
   Current smoker    
      Daily 2 567.01.713.80.5
      Other (a) 258.65.91.40.2
      Total current smoker2 824.81.515.10.4
   Ex-smoker5 440.80.829.20.5
   Never smoked10 388.10.555.70.5
   Total persons aged 18
    years and over (b)
18 654.20.0100.00.0

National Health Survey, 2017-18

 Estimate ('000)RSE of Estimate Proportion (%)MoE of Proportion 
Smoker status     
   Current smoker    
      Daily 2 568.11.713.80.5
      Other (a) 278.09.41.50.3
      Total current smoker2 840.31.615.20.5
   Ex-smoker5 585.61.329.90.8
   Never smoked10 227.30.854.80.9
   Total persons aged 18
    years and over (b)
18 656.20.0100.00.0

a. Includes current smoker weekly (at least once a week, but not daily) and current smoker less than weekly.
b. Discrepancy between 'Total persons aged 18 years and over' are due to random adjustments to avoid the release of confidential data.
 

30 When interpreting changes over time or differences between population groups (for example, between males and females), reliability of estimates should be taken into account. All comparisons in this publication were tested for statistical significance at the 95% level of confidence; for more information see Technical Note - Reliability of Estimates.

Classifications

31 Long-term health conditions reported by respondents in the NHS are presented using a classification originally developed for the 2001 NHS by the Family Medicine Research Centre, University of Sydney, in conjunction with the ABS. The classification is based on the 10th revision of the International Classification of Diseases (ICD) and is used for all years from 2001 to 2017-18.

32 Country of birth is classified to the Standard Australian Classification of Countries (cat. no. 1269.0).

33 Main language spoken at home is classified according to the Australian Standard Classification of Languages (cat. no. 1267.0).

34 Descriptions of data items such as Body Mass Index and the Kessler Psychological Distress Scale (K10) are included in the Glossary to this publication.

Confidentiality

35 The Census and Statistics Act, 1905 provides the authority for the ABS to collect statistical information, and requires that statistical output shall not be published or disseminated in a manner that is likely to enable the identification of a particular person or organisation. This requirement means that the ABS must take care and make assurances that any statistical information about individual respondents cannot be derived from published data.

36 To minimise the risk of identifying individuals in aggregate statistics, a technique known as perturbation is used to randomly adjust cell values. Perturbation involves a small random adjustment of the statistics and is considered the most satisfactory technique for avoiding the release of identifiable statistics while maximising the range of information that can be released. These adjustments have a negligible impact on the underlying pattern of the statistics. After perturbation, a given published cell value will be consistent across all tables. However, adding up cell values to derive a total will not necessarily give the same result as published totals. 

37 Perturbation has been applied to 2014–15 and 2017–18 data. Data from previous NHS presented in this publication have not been perturbed, but have been confidentialised if required using suppression of cells.

Rounding

38 Estimates presented in this publication have been rounded. 

39 Proportions presented in this publication are based on unrounded estimates. Calculations using rounded estimates may differ from those published.

Acknowledgements

40 ABS publications draw extensively on information provided freely by individuals, businesses, governments and other organisations. Their continued cooperation is very much appreciated; without it, the wide range of statistics published by the ABS would not be available. Information received by the ABS is treated in strict confidence as required by the Census and Statistics Act, 1905.

Products and services

41 Summary results from the NHS are available in spreadsheet form from the 'Data downloads' section in this release. The statistics presented are only a selection of the information collected. 

42 For users who wish to undertake more detailed analysis, a TableBuilder product for the 2017-18 NHS is expected to be available in the second quarter of 2019. TableBuilder is an online tool for creating tables from ABS survey data, where variables can be selected for cross-tabulation. It has been developed to complement the existing suite of ABS microdata products and services including Census TableBuilder and CURFs. Further information about ABS microdata, including conditions of use, is available via the Microdata section on the ABS website.

43 Customised tabulations are available on request. Subject to confidentiality and sampling variability constraints, tabulations can be produced from the survey incorporating data items, populations and geographic areas selected to meet individual requirements.

Related publications

44 Current publications and other products released by the ABS are listed on the ABS website. The ABS also issues a daily Release Advice on the website which details products to be released in the week ahead.

Appendix 1 - sample counts and estimates

Show all

Appendix 2 - physical measurements

Show all

Appendix 3 - self-reported height and weight

Show all

Appendix 4 - modelled estimates for small areas

Show all

Technical note - reliability of estimates

Reliability of estimates

1 Two types of error are possible in an estimate based on a sample survey: sampling error and non-sampling error. The sampling error is a measure of the variability that occurs by chance because a sample, rather than the entire population, is surveyed. Since the estimates in this publication are based on information obtained from occupants of a sample of dwellings they are subject to sampling variability; that is, they may differ from the figures that would have been produced if all dwellings had been included in the survey. One measure of the likely difference is given by the standard error (SE). There are about two chances in three that a sample estimate will differ by less than one SE from the figure that would have been obtained if all dwellings had been included, and about 19 chances in 20 that the difference will be less than two SEs. 

2 Another measure of the likely difference is the relative standard error (RSE), which is obtained by expressing the SE as a percentage of the estimate. The RSE is a useful measure in that it provides an immediate indication of the percentage errors likely to have occurred due to sampling, and thus avoids the need to refer also to the size of the estimate.

\(\Large R S E \%=\left(\frac{S E}{E s t i m a t e}\right) \times 100\)

3 RSEs for published estimates are supplied in Excel data tables, available via the Data downloads section.

4 The smaller the estimate the higher is the RSE. Very small estimates are subject to such high SEs (relative to the size of the estimate) as to detract seriously from their value for most reasonable uses. In the tables in this publication, only estimates with RSEs less than 25% are considered sufficiently reliable for most purposes. However, estimates with larger RSEs, between 25% and less than 50% have been included and are preceded by an asterisk (eg *3.4) to indicate they are subject to high SEs and should be used with caution. Estimates with RSEs of 50% or more are preceded with a double asterisk (eg**0.6). Such estimates are considered unreliable for most purposes.

5 The imprecision due to sampling variability, which is measured by the SE, should not be confused with inaccuracies that may occur because of imperfections in reporting by interviewers and respondents and errors made in coding and processing of data. Inaccuracies of this kind are referred to as the non-sampling error, and they may occur in any enumeration, whether it be in a full count or only a sample. In practice, the potential for non-sampling error adds to the uncertainty of the estimates caused by sampling variability. However, it is not possible to quantify the non-sampling error. 

Standard errors of proportions and percentages

6 Proportions and percentages formed from the ratio of two estimates are also subject to sampling errors. The size of the error depends on the accuracy of both the numerator and the denominator. For proportions where the denominator is an estimate of the number of persons in a group and the numerator is the number of persons in a sub-group of the denominator group, the formula to approximate the RSE is given below. The formula is only valid when x is a subset of y.
 
\(\Large R S E\left(\frac{x}{y}\right)=\sqrt{R S E(x)^{2-} R S E(y)^{2}}\)

Comparison of estimates

7 Published estimates may also be used to calculate the difference between two survey estimates. Such an estimate is subject to sampling error. The sampling error of the difference between two estimates depends on their SEs and the relationship (correlation) between them. An approximate SE of the difference between two estimates (x-y) may be calculated by the following formula:

\(\Large S E(x-y)=\sqrt{[S E(x)]^{2}+[S E(y)]^{2}}\)

8 While the above formula will be exact only for differences between separate and uncorrelated (unrelated) characteristics of sub-populations, it is expected that it will provide a reasonable approximation for all differences likely to be of interest in this publication.

9 Another measure is the Margin of Error (MOE), which describes the distance from the population value that the sample estimate is likely to be within, and is specified at a given level of confidence. Confidence levels typically used are 90%, 95% and 99%. For example, at the 95% confidence level the MOE indicates that there are about 19 chances in 20 that the estimate will differ by less than the specified MOE from the population value (the figure obtained if all dwellings had been enumerated). The 95% MOE is calculated as 1.96 multiplied by the SE.

10 The 95% MOE can also be calculated from the RSE by:

\(\Large MOE(y) \approx \frac{R S E(y) \times y}{100} \times 1.96\)

11 The MOEs in this publication are calculated at the 95% confidence level. This can easily be converted to a 90% confidence level by multiplying the MOE by:

\(\Large\frac{1.645}{1.96}\)

or to a 99% confidence level by multiplying by a factor of:

\(\Large\frac{2.576}{1.96}\)

12 A confidence interval expresses the sampling error as a range in which the population value is expected to lie at a given level of confidence. The confidence interval can easily be constructed from the MOE of the same level of confidence by taking the estimate plus or minus the MOE of the estimate.

Significance testing

13 For comparing estimates between surveys or between populations within a survey it is useful to determine whether apparent differences are 'real' differences between the corresponding population characteristics or simply the product of differences between the survey samples. One way to examine this is to determine whether the difference between the estimates is statistically significant. This is done by calculating the standard error of the difference between two estimates (x and y) and using that to calculate the test statistic using the formula below:

\(\Large \frac{|x-y|}{S E(x-y)}\)

where

\(\Large S E(y) \approx \frac{R S E(y) \times y}{100}\)

14 If the value of the statistic is greater than 1.96 then we may say there is good evidence of a statistically significant difference at 95% confidence levels between the two populations with respect to that characteristic. Otherwise, it cannot be stated with confidence that there is a real difference between the populations.

Glossary

Show all

Abbreviations

Show all

Back to top of the page