Scottish Health Survey - topic report: lung function

1. Background and Methods

Catherine Bromley, Lisa Rutherford, Jennifer Mindell

1.1 Introduction

Data on self-reported respiratory symptoms and doctor-diagnosed asthma in adults and children in Scotland were presented in the Scottish Health Survey (SHeS) 2010 annual report, using both symptoms and diagnoses data collected at interview.^[1] This report presents data from objective measurement of lung function in adults, measured by portable spirometers, using combined data from the 2008-2011 surveys. While the spirometers and protocols used in SHeS 2008-2011 were the same as those used in previous survey years, both the presentation and interpretation of spirometry data now differ in a number of significant ways, therefore no comparisons between 2008-2011 results and previous years are attempted here.^[2]

Three parameters of lung function are reported on: forced expiratory volume in one second (FEV₁), forced vital capacity (FVC), and the ratio of these measurements (FEV₁/FVC) (see Section 1.2.2). It should be borne in mind, when interpreting these results, that bronchodilators^[3] were not used by survey nurses in SHeS. The publication of these results was delayed while concerns about measurement error in some of the readings were examined. Further details on how these were identified and handled can be found in Section 1.4.4.

1.2 Lung Function

1.2.1 Overview

In healthy individuals, normal lung function changes over time and is dependent on age, sex, height, and ethnicity. Lung function increases twentyfold during the first 10 years of life, with rapid growth continuing through adolescence.^[4] While both lung volume and forced expiratory volumes increase during childhood and teenage years, they do not do so at the same rate.^[5] In childhood, FVC grows faster than FEV₁, so the FEV₁/FVC ratio falls. In adolescence, these trends are reversed,^[5] and both (FVC and FEV₁) then decrease with age in adulthood. Lung and airway function are largely determined by foetal development and early life events during infancy.^[6] Those on lower centiles^[7] in childhood, particularly those with intra-uterine growth retardation (IUGR), tend to have lower lung function as they grow older.^[8][9] Children born with IUGR are more likely to have lung function below the 10^th centile throughout childhood as well as an increased risk of respiratory disease as adults.^[10]

Lung functionpeaks among young adults,^[11][12] with highest values found at age 23 in men and 22in women before declining steadily with age.^[13][4] Age-related changes in FEV₁ and FVC result in age-related changes in the FEV₁/FVC ratio: the lower limit of normal for the FEV₁/FVC ratio has been considered to be around 0.7 at age 45, but drops below this after this age.^[14] Because lung function tracks throughout life, maternal smoking during pregnancy, active or passive smoking in childhood and adolescence, and other factors that reduce lung function in infancy and childhood^[6] still have an effect in adulthood, as does active and passive smoking in adulthood.^[15]

1.2.2 Changes in interpretation of lung function

To distinguish between normal and abnormal results for a specific individual, lung function tests need to be adjusted for age, sex, height and ethnic group. Advances in computing now mean that complex computation of norms ('predicted values') for a healthy person of a given age, sex, height and ethnic group can be provided instead of the previous, simpler, approach of using broad categories combining these variables within ranges. Thus it is now possible to present a person's results as a percentage of the predicted value for that individual, as has been used in the past for diagnostic purposes.

The distribution of measurements in healthy individuals varies with age, and is wider in older people. Previously, abnormal FEV₁ or FVC was defined as less than 80% of the predicted value, while a fixed threshold of 0.7 was used to define an abnormal FEV₁/FVC ratio. However, because of the changes in lung function that occur in healthy people with age, these rigid cut-offs overestimate abnormality in older people while underestimating it among younger adults.^[5][52] Many experts now consider it to be more valuable to present the results as centiles^[7] independent of age, sex, height or ethnicity. A measure of abnormality can then be defined based on such centiles instead of the absolute percentage of the predicted values for individual participants. For example, those whose lung function, as measured by spirometry, is below the 5^th centile for the 'normal' population can be defined as having poor lung function.

In recent years there have also been marked changes concerning the appropriate equations to derive predicted values for use in studies of lung function. In the past, the European Coal and Steel Community reference equations^[16][17] were used extensively, but these have been criticised^[18] and have been shown to be inaccurate, particularly for assessing women's lung function.^[19] Stanojevic 2009 all-ages equations^[20] have been used in the analysis presented in this report. Prepared by an international collaboration, these equations are based on data from healthy individuals aged four to 80 years,^[11] subsequently extended to include children aged 3 years.^[20] The equations cover the widest age-range of any available equations; provide smoothly changing curves to describe the transition between childhood and adulthood, instead of the discontinuities seen when separate equations are used for children and adults; and incorporate the relationship between height and age in a biologically plausible manner. These reference equations confirm that the range of normal values varies markedly by age.^[20]

1.3 Respiratory Disease

1.3.1 Asthma

Asthma may be present if repeated peak flow measurements show significant variation by time of day or from day to day, or if FEV₁ increases by at least 400ml in response to bronchodilators or a two week course of oral steroids.^[21] While asthma is defined in relation to reversibility of airways obstruction, adults with asthma can develop irreversible airways obstruction.^[22] Asthma symptoms in adults do not correlate well with lung function measurements.^[23] Lung function in adults with asthma is affected by parental asthma, repeated early-life wheezing, and both early life and current passive smoking, as well as active smoking.^[17][24]

Diagnosis is based on a careful history that reveals a characteristic pattern of symptoms and signs with no alternative explanation for them. Spirometry is the preferred initial test to assess the presence and severity of airflow obstruction.^[25]

1.3.2 Chronic Obstructive Pulmonary Disease (COPD)

Definition, aetiology and prevalence

The World Health Organization (WHO) defines Chronic Obstructive Pulmonary Disease (COPD) as 'a lung disease characterised by chronic obstruction of lung airflow that interferes with normal breathing and is not fully reversible.'^[26] COPD is characterised by airflow obstruction that usually progresses and, unlike asthma, does not change markedly over several months and is not fully reversible by bronchodilators.^[22] Clinically significant COPD is not present if FEV₁ and the FEV₁/FVC ratio return to normal with drug therapy.^[25]

Seldom diagnosed before the age of 50, COPD comprises the spectrum of diseases previously called 'chronic bronchitis' and 'emphysema.' Symptoms of 'chronic bronchitis' include a productive cough, producing phlegm most mornings for at least three months of the year; shortness of breath (dyspnoea); and fatigue leading to exercise intolerance.^[27] Exercise intolerance can be present in patients with only mild disease although the extent of exercise intolerance is generally in proportion to the severity of the disease. COPD is, however, both preventable and treatable.^[28]

Smoking, the most important cause of COPD, contributes to both the development and progression of the disease as exposure to tobacco smoke damages both the lung tissue and the airways.^[29] COPD is also caused by alpha-1 antitrypsin deficiency, a genetic defect; exposure to dust, chemicals and gases in the workplace; and other environmental exposures, such as particulate pollution.^[30][31][32] Smoking cessation can slow decline in lung function in COPD.^[33]

COPD affects about 9% of adults across Europe;^[34] and it is predicted that it will become the third most common cause of death worldwide by 2030.^[35] Figures presented in Audit Scotland's 2007 report on managing long-term conditions suggested that around 100,000 people in Scotland were thought to have COPD, with prevalence predicted to increase by one-third between 2007 and 2027.^[36] In the 2008-2011 period, death rates from COPD ranged between 34.1 and 41.0 per 100,000 men, and 30.7 to 32.3 per 100,000 women in Scotland.^[37] Between 1979 and 2012, the annual mortality rate for COPD in Scotland in men fell by 50% but increased in women by 76%, with the ratio of male to female death rates falling from 4.0 to 1.2 in the same period.^[40] COPD is the only cause of death that is rising in Scotland, accounting for around 4,500 deaths annually.^[38]

The costs to the health service resulting from COPD are also considerable. In 2010/11, 14.9 patients per 1,000 men and 17.6 per 1,000 women consulted a GP or practice nurse at least once about COPD.^[39] It results in more than 122,000 bed days annually and costs NHS Scotland £100 million per year.^[41] The prevalence of COPD, the likelihood of consulting healthcare professionals in primary care, and the risk of emergency admissions all increase with deprivation.^[39][40]39, Both the differing patterns by sex and the link with deprivation are primarily associated with differential smoking rates over the previous 30-40 years, as well as with exposure to air pollution.

Despite the importance of COPD and the increasing policy focus on long-term conditions, few policies relate specifically to COPD. Most consider generic approaches to improving management of long-term conditions,^[41] including improving palliative care.^[42][43] One of the NHS HEAT (Health improvement, Efficiency/governance, Access, Treatment) targets for 2008/09 to 2010/11 was for agreed reductions in hospital admissions for four chronic conditions, including COPD.^[44] COPD was one of the exemplar conditions in Audit Scotland's report on managing long-term conditions. It recommended increasing community care to reduce admissions, outpatient visits, and GP consultations.^[39] A UK-wide COPD audit of specialist care, co-ordinated by the Royal College of Physicians in 2007,^[45] was followed by peer-visits. The repeat audit in 2010 found small changes, with no major differences between the intervention and control units.^[46]

The Global Initiative for Chronic Lung Disease (GOLD) aims to raise awareness of COPD.^[31] Delays in diagnosis of COPD can occur if smokers consider their productive cough to be 'normal', particularly if friends and family who smoke have similar symptoms.^[47] Scottish quality standards for organising COPD services and to improve the identification and treatment of people with COPD were published in 2010.^[41] It was recommended that spirometry is used to confirm the presence of chronic airways obstruction as part of making the diagnosis, which is made on clinical judgement based on the history, physical findings and the spirometry findings.^[41]

Spirometric findings to diagnose COPD

Until recently, airflow obstruction was generally defined as an FEV₁/FVC ratio of the post-bronchodilator spirometry of less than 0.70.^[48] However, as explained in Section 1.2.2, this fixed ratio causes up to 50% over-diagnosis in people aged over 45 years.^[15] The 2011 five year major revision of the GOLD strategy continued to recommend use of fixed thresholds,^[49] as does the 2014 edition,^[50] although it acknowledges this problem, and recommends the use of other clinical symptoms and history to make a diagnosis, while stating that spirometry is "required to make a confident diagnosis of COPD." GOLD has also defined the severity of COPD by the reduction in forced expiratory volume in one second (FEV₁), relative to the value predicted for age, height, and sex,^[23][51] although the National Institute for Health and Clinical Excellence (NICE) restricts the term 'mild COPD' to those with symptoms of COPD as well as a reduced FEV₁/FVC ratio.^[24]

Increased recognition of the age-related changes in lung function^[52] (referred to in Section 1.2.2) above has resulted in changes to the interpretation of spirometric assessment of COPD,^[51] with calls for revised internationally accepted definitions.^[15][52] Recent recommendations to avoid age-related distortions from a fixed threshold are that the definitions of abnormal lung function should be based on spirometry falling below the 5^th centile of predicted (referred to as the 'lower limit of normal').^[54]

1.4 Methods and Definitions

1.4.1 Procedures for measuring lung function on SHeS

Nurses conducted spirometry during their visit to the participants. Adult participants were excluded from the lung function measurement if they:

• were pregnant;
• had had abdominal or chest surgery in the last three weeks;
• had been admitted to hospital with a heart complaint in the preceding six weeks;
• had had eye surgery in the preceding 4 weeks;
• had been admitted to hospital with a heart complaint in the preceding month; or
• had a tracheostomy.

While all remaining adults were eligible for spirometry, the nurse had flexibility to use their clinical judgement if they felt a participant was too ill to be asked, or had a medical condition which made spirometry inadvisable.

The equipment used was a Vitalograph Escort spirometer. The calibration of the spirometer was checked with a 1 litre calibration syringe in the nurse's home each day before being taken to a participant's home for use in the survey.

No bronchodilators were given to participants by the survey nurses. However, participants who normally took a puff of their inhaler before strenuous exercise were allowed to do so before the spirometry measurements.

The nurse began by explaining the purpose of the test, how to use the spirometer, and the importance of blowing as hard, and for as long, as possible to obtain an accurate measurement. Participants were also told that they would not be provided with an interpretation of the results during the interview as assessing lung function depends on age, sex and height and the diagnosis of abnormal lung function depends on their clinical history and on measurements taken on more than one occasion.

The nurse then demonstrated the procedure to the participant. To perform the procedure, the nurse instructed the participant to breathe in as deeply as possible, place the lips (not the teeth) firmly round the mouthpiece, and then immediately blow out the air as hard and as fast as they could, and to keep blowing until there was no more air in their lungs. The blow needed to be at least 3 seconds in length and not interrupted by coughing, laughing or leakage of air. The torso was to remain in an upright position throughout the blow, not hunched over at the end.

Spirometry was performed standing up, except where the participant was chairbound. The participant was asked to loosen tight clothing, to allow a larger inspiration. The participant was asked to perform at least one practice blow before the mouthpiece was attached to the spirometer.

The nurse then asked the participant to take as deep a breath as possible, keeping the spirometer away from their mouth, and then to hold the mouthpiece with their lips and seal their lips around it so that air did not escape while they are blowing. After instructing the participant to start blowing, the nurse continued to encourage the participant by saying "keep going, keep going, keep going..." to get the maximum expiration possible. The nurse then took the spirometer, recorded the results, and reset the spirometer. This was repeated until, the participant had made five attempts, or the nurse deemed the participant was too tired to continue.

A technically unsatisfactory blow was noted when any of the following were observed by the nurse:

• an unsatisfactory start, e.g. excessive hesitation or a 'false start';
• laughing or coughing, other than in the final second of the blow;
• holding the breath in;
• a leak in the system or around the mouthpiece;
• an obstructed mouthpiece (e.g. by the tongue or false teeth); or
• a recording of 0.00 for FEV₁, indicating that the test was not conducted properly.

The full spirometry protocol is available in Annex to this report.

1.4.2 Definitions - Lung function parameters

Definitions of the measurements used in this chapter are listed in Table 1A below.^[53] The measurements do not refer to normal breathing, but to a forced manoeuvre where the lungs are filled as deeply as possible and the air is then forced out as fast and as hard as possible, and the manoeuvre continues until all air is expelled.

Table 1A Lung function test parameters

Test	Abbreviation	Measurement unita	Definition	Lay explanation
Forced Vital Capacity	FVC	litres	The total volume of air that can forcibly be blown out after a full inspiration	This indicates the 'size' of the lungs.
Forced Expiratory Volume in 1 Second	FEV1	litres	The volume of air that can be blown out in one second during a forced manoeuvre	This measures how easily an individual can breathe out. It depends on how wide (dilated) the airways are.
FEV1 as a proportion of FVCb	FEV1 / FVC	proportion or ratio	The ratio of FEV1 to FVC.	This measures the proportion of the air in the lungs that an individual can breathe out in the first second.

^a Although this was the unit of the measurement, these are not the units presented in this report, as explained in Section 1.2.3 below.

^b This value was derived during the post-fieldwork data processing, using the FEV₁ and FVC values entered by the nurses. The spirometers calculated FEV₁/FVC ratio, but nurses were not required to input this.

The pattern of results for the three different measures is particularly important in diagnosing and monitoring disease. For diseases where there is airflow obstruction, FEV₁ and the FEV₁/FVC ratio are low but FVC is relatively unaffected (for example asthma and COPD). For restrictive diseases, such as fibrosing alveolitis, lung volume is reduced: although both FEV₁ and FVC are reduced, generally by similar amounts, the FEV₁/FVC ratio is relatively unaffected.

1.4.3 Measurement quality

Unlike spirometry performed in a respiratory clinic, survey nurses were not spirometry specialists, and measurements were taken in participants' homes. Each nurse conducted relatively few spirometry sessions, and while some worked consistently throughout the year, some worked less regularly, with gaps of two or three months between monthly assignments (which were 6-7 households on average). Each nurse involved in SHeS underwent training using spirometers prior to starting work on the survey. The equipment used in SHeS did not provide any formal assessment of spirometry quality. While some range checks to alert nurses to unusually high or low figures were built into the computer assisted interviewing programme (CAPI) used by the nurses, manual entry of data meant the chance of human error was not completely eliminated (see more on this below). Nurses were, however, able to comment on data if they felt the results did not accurately represent the participant's lung function, for example, because of poor quality blows.

When interpreting the findings in this chapter, it should also be borne in mind that conducting spirometry in people's own homes has other differences from an outpatient clinic or respiratory laboratory. For example, not all individuals have a suitable chair available to provide optimum support for participants to sit up straight with their feet on the floor. It is also harder for staff to be insistent about making the considerable effort required to provide a sufficient number of good quality blows in a home rather than clinic setting.

1.4.4 Measurement error and data imputation

When the 2011 SHeS report was originally being prepared some concerns with the SHeS FEV₁ data were identified, largely due to the results being very different to those found in the 2010 Health Survey for England (HSE) (Scotland's results were notably poorer than England's, and worse than any previous respiratory epidemiology had suggested). Since the two surveys used quite different methods for measuring lung function, a number of investigations had to be carried out to determine whether the difference seen was a methodological artefact, a data problem, or a true population finding. The two biggest methodological differences were that the HSE spirometers provided feedback to nurses and participants on the quality of the blows performed while measurements were being conducted, and that results were transmitted directly to the nurses' laptops, eliminating the potential for data entry errors.

It is certainly the case that HSE procedures resulted in a far lower proportion of usable spirometry readings being collected: 65% of men and 67% of women who took part in the nurse visit in HSE yielded usable spirometry data.^[54] However, weighting was applied to the HSE results to correct for biases that might have been introduced due to the large number of excluded cases, which meant that differences in the profile of participants in the two surveys, on which the results were based, were actually quite small. Where they did exist, this was largely due to there being underlying differences between the two populations. For example, the prevalence of smoking among those who provided valid spirometry was higher in Scotland than England for most age groups, but this was caused by a higher overall prevalence of smoking in Scotland than England (data not shown).

The fact that the SHeS nurses had to enter the spriometry results into their laptops manually also opens up the possibility that data entry errors were higher in Scotland than in England, where the equivalent data were transmitted electronically. A specific data entry error was highlighted during SHeS fieldwork wherein a nurse was found to have entered the values for a parameter labelled in the spirometer output window as "FEV1%" (which was in fact the ratio of FEV₁ to FVC) rather than the FEV₁ values. As Table 1A above notes, the FEV₁/FVC figures presented in this report were derived from the individual FEV₁ and FVC values in the dataset, nurses were not required to input that value directly. On the basis of this known error potential, further investigations assessed the internal validity of the data to see whether this had been more widespread than a handful of isolated cases. Plotting FEV₁ results by peak flow (PF), which should show a good degree of correspondence, helped to identify a distinct subset of outlier cases whose FEV₁ values did not vary as PF increased (all these cases had FEV₁ values below 1.0L (data not shown).

This analysis confirmed that there was likely to have been a more widespread data error entry than had been originally thought, with a potential 444 cases identified as having potentially suspect results (all those with FEV₁ results below 1.0L and FVC results above 2.5L) over the four year period. Analysis showed that the errors were made by a small number of nurses only. As noted above, the data entry error resulted in nurses entering a value that was a ratio of FEV₁ to FVC, rather than the FEV₁ value itself. This meant that such cases could have a correct FEV₁ value imputed, by multiplying the value originally entered as FEV₁ by the FVC. However, to exclude the possibility that some of these cases were genuine poor FEV₁ readings, a further subset of participants was identified whose PF results were more than two standard deviations below the mean for their sex (61 cases). This left 383 cases where there was a strong likelihood that the FEV₁ value originally entered was in fact FEV₁/FVC. Imputing FEV₁ for these 383 cases resulted in the correlation between FEV₁ and PF being much higher, and following the expected pattern. For these cases it is the imputed data that has been included in the analysis presented in Chapter 2.

No imputation process is perfect, and it is certainly not the case that every possible data entry error or other measurement error will have been identified and resolved by this approach (for example, no trimming of extreme FEV₁, FVC or PF values was applied). This method of resolving the implausible FEV₁ outliers was deemed preferable to other options, such as excluding the cases altogether, or leaving the original results unadjusted. However, both the original and imputed FEV₁ results are available in the public dataset should secondary analysts wish to pursue other options.

1.5 Response to Lung Function Measurement

Valid spirometry was achieved in 96% of adults who had a nurse visit. Just 2% of adults refused spirometry, while a further 2% were judged to be ineligible (see Section 1.4.1). The majority of adults provided five technically satisfactory (and therefore usable) blows (86% of men and 81% of women), and a further 11% of men and 14% of women provided at least 1 technically satisfactory blow. The proportion deemed ineligible was highest in women aged 25-34 (6%), for whom pregnancy was an exclusion criteria, and among those aged 75 and over (5% of men and 4% of women). The latter of these partly explains the lower proportion of participants aged 75 and over producing adequate quality spirometry (90% of both men and women aged 75 and over). In contrast, 95%-99% of adults aged 16-64 provided usable spirometry readings. The proportion of adults able to provide all five technically satisfactory blows also declined from age 35 onwards.

Not surprisingly, the proportion of participants with valid spirometry measurements was lower in older adults. Self-reported COPD or asthma, and smoking status did not affect the proportion of individuals of that age and sex who provided adequate spirometry (data not shown).

Overall, 1,776 men and 2,273 women (96% and 95% respectively of those having a nurse visit) provided at least one technically satisfactory spirometry measurement (of which, 383 cases had values for FEV₁ imputed to correct for a data entry error, as detailed in Section 1.4.4). However, the results presented in this report use data from a European reference population (as detailed in full in Chapter 2) adjusted for age, height and sex, which are only valid for the white (Caucasian) population. Therefore, the tables presented in Chapter 2 are based on the 1,734 men and 2,216 women for whom the reference equations could be applied.

Table 1.1