MJA
MJA

An algorithmic approach to diagnosing asthma in older patients in general practice

Richard E Ruffin, David H Wilson, Sarah L Appleton and Robert J Adams
Med J Aust 2005; 183 (1): S38. || doi: 10.5694/j.1326-5377.2005.tb06917.x
Published online: 4 July 2005

Abstract

What we need to know

  • How effective would an algorithm be in helping general practitioners diagnose asthma?

  • What proportion of older people with undiagnosed asthma fail to recognise symptoms?

  • What proportion of the population believe asthma does not occur in the older population?

  • What systems or supports do GPs need to diagnose asthma more effectively?

What we need to do

  • Work on developing a gold standard for asthma diagnosis.

  • Develop prototype algorithms for general practice discussion.

  • Conduct a general practice study to assess the effectiveness of an algorithm.

  • In conjunction with GPs, develop a pilot program to increase awareness of the current asthma problem.

  • Conduct focus-group research to identify why some people do not believe they can develop asthma for the first time in adult life.

  • Conduct focus-group research to identify why some adults do not attribute asthma symptoms to asthma.

  • Conduct focus groups with GPs to identify what support is needed to diagnose asthma more effectively.

  • Consult with all stakeholders before an intervention is used.

  • Evaluate any interventions used.

Arecent study has estimated that the prevalence of undiagnosed asthma in the adult population is 2.3%,1 and that this prevalence almost doubles in those aged over 65 years.2 Given that patients whose asthma is undertreated have worse outcomes, recognition of people with undiagnosed and untreated asthma is important.3 Although there is no Level I evidence4 to support the idea, an algorithm to aid the early diagnosis and treatment of asthma in older people may improve their respiratory outcomes. For the purpose of this discussion, we define an algorithm according to Stedman’s medical dictionary as a “step-by-step [written] protocol for management of a health care problem”.5 It is important to consider the issues associated with making an asthma diagnosis overall, and specifically in older people in whom other issues of ageing play a part, and whether these can be accommodated in an algorithm.

It is difficult to define asthma accurately. The definition of asthma by the Global Initiative for Asthma6 includes inherent ambiguities (eg, “episodes are usually associated with widespread but variable airflow obstruction that is often reversible . . .”). Another issue is the lack of a gold standard for diagnosing asthma, which has implications for the sensitivity and specificity of the algorithm. However, this will be refined in future algorithms, and does not prevent us from using agreed best practice now.

International guidelines for the diagnosis and management of asthma suggest that a significant change in forced expiratory volume in 1 second (FEV1) after use of a bronchodilator is indicative of asthma. However, the degree of reversibility of airflow restriction that is considered significant varies between guidelines, from 12% to 15% of the baseline value.6,7 In addition, there is inconsistency between guidelines as to whether the post-bronchodilator response should be a percentage of baseline or of predicted FEV1. There is also controversy surrounding the diagnostic value of the response to bronchodilator.8 The UK National Institute for Clinical Excellence, in its recent guideline on chronic obstructive pulmonary disease (COPD), recommended a 400 mL increase in FEV1 in response to bronchodilator as diagnostic of asthma.9 Given the relative diagnostic uncertainty, the utility of an algorithm may be in providing a pathway for an approach to decision-making.

Several complex and possible contributory factors need to be considered in understanding why asthma is underdiagnosed in older people, and an algorithm may not be able to address this problem. Altered perceptions of dyspnoea have been described in older people,10 and patients under-report symptoms to their doctor.11 However, the proportions who dismiss their asthma symptoms as part of the normal ageing process, or deny that they have symptoms, or experience classical as opposed to non-classical symptoms of asthma, is unknown. These are barriers to presentation and not to diagnosis, and should be dealt with in health education campaigns.

Probably the most relevant issue to consider is the acceptability and utility to general practitioners of yet another diagnostic algorithm. After consultation with GPs to determine what would be useful, feasible and applicable in general practice, given their brief interactions with patients of all types, it was evident that GPs would use an algorithm as a reference tool rather than a day-to-day tool.2 If an algorithm establishes improved decision-making in general practice for asthma diagnosis, it has served its purpose and does not need to be used on a daily basis.

GPs also related that they think in terms of “bundles” of information rather than in a linear (algorithmic) way. An algorithm developed from these consultations that attempted to accommodate this point made by GPs is shown in the Box.

Of equal importance, GPs were unaware of the reported level of underdiagnosis of asthma and felt strongly that being aware of this would make a difference to their approach. This consensus view is consistent with the National Asthma Campaign’s “Could it be asthma?”, an awareness program promoted to the community and health professionals. The program was associated with an increase in self-reported doctor-diagnosed asthma in adults from 5.6% in 1987 to 8.0% in 1990.12 Information technology advances may make algorithms or pathways more relevant and acceptable to GPs in the future. Clinical technology advances, supported by evidence that they work in the relevant clinical setting, will need regular review and appropriate incorporation into practice.

Is it critical that an algorithm should be able to separate asthma from COPD? While treatment may seem to be similar for both diseases, at milder levels of disease there are differences in treatments and acute outcomes, and at all levels there are differences for prognosis. Thus, doctors should distinguish between asthma and other respiratory conditions.

It is important to consider the required elements of an algorithm, including symptoms and tests (Box). There have been efforts to determine the positive predictive values of symptoms, with diagnosis by a doctor as the gold standard. One study reported that wheezing associated with rest dyspnoea or nocturnal dyspnoea showed positive predictive values of 42% and 39%, respectively, for diagnosing asthma.13 Thus, limitations of data in the literature must be considered in developing the algorithm. The medical history can provide an indication for the diagnosis at the level of more likely or less likely to support the diagnosis. Physical examination is less helpful and usually supports a “rule out” of more sinister diseases. Tests such as chest x-ray are important to rule out other diseases, and spirometry has limitations associated with access, acceptability of definitions of “reversibility”, repeatability of bronchodilator response and quality. Determining the place of a therapeutic trial or other objective outcomes, such as functional limitations or activities of daily living, needs further work and evidence.

Online responses are no longer available. Please refer to our instructions for authors page for more information.