Measuring Disease Burden: How and Why

Policy Matters, Understanding CFS | 13. Oct, 2011 by Guest Contributor | 14 Comments

by Andrew Kewley

man-carrying-a-heavy-dollar-symbolThe burden of disease refers to the burden that ill health and risk factors place on society. Measures of this burden include prevalence, mortality, life expectancy, economic costs, hospitalisation rates as well as specific measures of quality of life and disability.

Although no one wishes to diminish the personal suffering of those suffering from ill health, these measures can be used to compare levels of public health spending or levels of research funding and scientific interest. These comparisons can be useful to identify any large inequalities in funding that could lead to an increased societal burden of disease. I strongly suspect that there are substantial inequalities in the level of funding for myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) because the overall burden of this disease has not yet been shown in a manner comparable to other diseases. I propose that the disease burden of ME/CFS should be studied further, so that the awareness of its impact can be increased.

The Disease Burden of ME/CFS
The primary studies attempting to show the burden of disease of ME/CFS have been prevalence studies. In particular, population based studies which use random sampling methods to call households in particular localities and screen individuals for their likelihood of having a specific disease. These studies tend to have less biases than other studies which try to estimate prevalence or incidence based on the rate at which patients visit primary care physicians. In the USA, there have been three large population based studies which estimated the prevalence of ME/CFS. The first of which was a study in Chicago, Illinois [1] which estimated a prevalence of 0.42%, the second was a study in Wichita, Kansas [2] which had an estimated prevalence of 0.24%, both studies utilised the 1994 Fukuda/CDC case definition for Chronic Fatigue Syndrome. A third study conducted in Georgia [3] had a much higher estimate of 2.54%, due to using the more inclusive CDC empirical criteria.

These population based studies also permitted the economic costs of this disease to be estimated. The first study was based on the Wichita cohort and estimated the annual cost of lost productivity to be $9.1 billion in the USA (2004 est.) [4].  A second study estimated the economic costs in two Chicago samples. The total annual economic costs for the USA were estimated to be $18.7 billion dollars for the population sample and $24.0 billion dollars for the tertiary sample (2008 est.) [5]. A third study was based on the Georgia population study and estimated annual economic costs in the USA of $51 billion dollars (2011 est.) [6]. This larger estimate reflects the higher incidence, but lower individual impact of the disease. Given the demonstrated economic burden of this disease, it is disappointing that how limited the level of scientific interest and research funding has been.

Measuring Quality of Life with ME/CFS
There have been a number of studies on ME/CFS using standardised measures of quality of life, in addition to studies using specific measures which have demonstrated a range of symptoms and substantial impairment among sufferers of this disease. Unfortunately, due to substantial variations in methodology, these findings have not been directly comparable in terms of measuring the societal burden of disease. However in general, most studies have reported substantial reductions in the number of hours employed [7]. This is due to the fact that many patients experience severe increases in symptoms and impairment when low or modest activity limits are exceeded [8]. Non-pharmacological treatments, such as cognitive behavioural therapy have also targeted the low activity levels of ME/CFS patients, however no studies have yet demonstrated objective increases in activity levels compared to controls [9], lending more support to the idea that patients have low physical activity limits [10].

The Disability Adjusted Life Year (DALY)


From PLoS ONE, reference 15

Due to the limitations of other measures of disease burden, the Disability Adjusted Life Year was developed by the World Health Organisation (WHO) for the 1990 burden of disease study [11]. This metric attempts to capture the overall disease burden due to early death or disability over time. Put simply, it is the sum of years that individuals in a particular population will lose due to early death, or an equivalent amount of time lost due to disability. This measure has been used by the WHO and other public health organisations to estimate global and national disease burden due to disease and potential risks of injuries [12]. Several comparisons have also been made between funding levels for specific diseases by the National Institutes of Health in the USA [13] and measures of disease burden [14, 15]. In particular, only DALYs were found to be highly correlated with levels of research funding [14, 15].

Unfortunately, ME/CFS has not been included in the WHO estimates of disease burden and hence any potential inequalities in funding have not yet been shown. I suggest that once an accurate DALY calculation is published, it will a substantial inequality in research funding.

Calculating DALYs for ME/CFS
The formula for the Disability Adjusted Life Year is inherently incidence based, since this method allows predictions based on changing demographics over time. This method also allows discounting for time preferences, to reflect that fact that people tend to favour time in the near future over the distant future.

The key variables for the calculation are:

  • Rates and age of premature mortality
  • Incidence and duration of disease
  • Disability weighting due to the morbidity impact of the disease
  • Discounting factors for age weighting and time preference (for comparison with WHO data)

ME/CFS is associated with factors such as reduced activity levels and this is associated with reduced life expectancy. However ME/CFS is rarely reported as a cause of death, making the first variable difficult to calculate. In contrast, the impact of the disease due to the morbidity is substantial and can be determined.

There have only been a few studies which have attempted to directly measure the incidence of ME/CFS, however this is typical of many diseases. It is common practise for incidence data to be extrapolated from data on prevalence and duration, using Dismod software [16] or other methods. If the prevalence is constant, then the burden can be computed in a straightforward manner without incidence data. Prevalence data can be provided by the previously mentioned population based studies [1-3]. The duration of the disease can be estimated from longitudinal studies. These include the CDC longitudinal surveillance study, which enabled estimates on the cumulative probability of recovery [17] and the 10 year follow-up to the Chicago population study [18, 19]. The Chicago follow-up study found that over the 10 year period, there was a recovery rate of 33% of those who were re-evaluated, but overall incidence remained constant [18, 19]. The CDC surveillance study found that although 48.1% had reported temporary recovery/remission at some period within 10 years, approximately 33% of those who continued to participate in the study later reported a return of their illness [17]. Which means that approximately 30% of patients had recovered after 10 years.  A systematic review of recovery rates for ME/CFS reported low median recovery rates of 5%, with a range of 0-31% [20]. The reason why the median rates reported in that review are lower than the longitudinal studies is likely due to the shorter follow-up intervals of the studies considered.

WHO-Global-burden-coverThe disability weightings for the original WHO disability studies were based on the consensus of a panel of experts using a “Person trade off” approach [21]. However since then, studies have utilised more empirical and systematic methods of estimating disability weights and any changes over time.  Estimations of disability weights can utilise standardised quality of life questionnaires such as the SF-36 which was utilised in the Chicago population studies [1, 18, 19]. Other studies on disability weights have conducted systematic reviews of the relevant literature and performed Monte Carlo sensitivity analyses to estimate the uncertainties of the disability weights. These analysis can consider a range of symptoms and impairments, their associated weightings and the probability of a given patient having that symptom, as well as any changes over time.

Discounting for age and time preference and other limitations of the DALY method, have been criticised in the literature [22]. However DALYs can be computed without discounting in addition to using the same discounting methods as the WHO burden of disease studies. In terms of the other criticisms, the DALY metric is still the most used and therefore necessary if comparisons are to be made.

Further Considerations
Recently there has been increased debate about the utility of different diagnostic criteria. With much of the debate focusing on the specificity and sensitivity of the different criteria, including the 1994 Fukuda/CDC Case definition [23], the Canadian Clinical Case definition [24], the CDC empirical criteria [25] and the recently published Myalgic Encephalomyelitis: International Consensus Criteria [26]. Naturally the use of different case definitions will lead to different estimates of prevalence, incidence and the burden of disease. However, despite any ambiguities about the most suitable definition, I believe it is worthwhile to estimate the disease burden based on existing data which is based on the 1994 Fukuda/CDC definition. This definition has been used in the majority of studies so far and should allow a reasonable estimate. Once population based incidence/prevalence studies are conducted for other case definitions, new estimates of disease burden can be performed and the overall disease burden compared.

I expect that a formal measurement of the disease burden using the DALY method will lead to increased awareness among the medical and scientific community of the burden of this disease and the importance of increasing the level of scientific research.

Andrew Kewley lives in Adelaide, Australia and closely follows the evolving research literature on ME/CFS.

[1] Jason L et al. A community-based study of CFS. Arch Intern Med. 1999, 159:2129-2137.

[2] Reyes M et al. Prevalence and incidence of CFS in Wichita, Kansas. Arch Intern Med. 2003, 163:1530-1536.

[3] Reeves WC et al. Prevalence of chronic fatigue syndrome in metropolitan, urban, and rural Georgia. Population Health Metrics 2007, 5:5.

[4] The economic impact of CFS. K Reynolds et al. Cost Eff Resour Alloc. 2004, 2: 4.

[5] Jason L et al. The economic impact of ME/CFS: Individual and societal costs. Dynamic Medicine. 2008, 7:6.

[6] Lin J et al. The economic impact of CFS in Georgia: direct and indirect costs. Cost Effectiveness and Resource Allocation. 2011, 9:1.

[7] Ross S et al. Systematic review of the current literature related to disability and CFS. Evidence Reports/Technology Assessments, No. 66, December 2002.

[8] Jason L et al. The energy envelope theory and ME/CFS. AAOHN J. 2008, 56(5):189-95.

[9] Wiborg J et al. How does cognitive behaviour therapy reduce fatigue in patients with CFS? The role of physical activity. Psychol Med. 2010, 40(8):1281-7.

[10] Brown M et al. The role of changes in activity as a function of perceived available and expended energy in nonpharmacological treatment outcomes for ME/CFS.  J Clin Psychol. 2011, 67(3):253-60.

[11] Murray C. Quantifying the burden of disease: the technical basis for disability-adjusted life years. Bulletin of the World Health Organization. 1994, 72(3):429-45.

[12] World Health Organisation Global burden of disease website:

[13] Estimates of Funding for Various Research, Condition, and Disease Categories (RCDC).

[14] Gross C et al. The relation between funding by the National Institutes of Health and the burden of disease. New England Journal of Medicine. 1999; 340:1881-1887

[15] Gillum L et al. NIH disease funding levels and burden of disease. PLoS ONE. 2011, 6(2): e16837.

[16] Dismod II, provided by the WHO:

[17] Reyes M et al. CFS progression and self-defined recovery: Evidence from the CDC surveillance system. Journal of Chronic Fatigue Syndrome. 1998, 5:1, 17 —27.

[18] Jason L et al. CFS prevalence and risk factors over time. Journal of Health Psychology. 2011, 16(3):445-56

[19] Jason L et al. A natural history study of CFS. Rehabil Psychol. 2011 Feb;56(1):32-42.

[20] Cairns R, Hotopf M. A systematic review describing the prognosis of CFS. Occupational Medicine (Lond). 2005, 55(1):20-31.

[21] Murray C, Acharya A. Understanding  DALYs. Journal of Health Economics. 1997, 16(6):703-30.

[22] Mont D. Measuring health and disability. The Lancet. Volume 369, Issue 9573.

[23] Fukuda K et al. The Chronic fatigue syndrome: A comprehensive approach to its definition and study. Annals of Internal Medicine. 1994;121:953-959.

[24] Carruthers B et al. ME/CFS: Clinical working case definition, diagnostic and treatment protocols. Journal of Chronic Fatigue Syndrome, 2003, 11:1, 7 —115

[25] Reeves et al. Chronic fatigue syndrome – A clinically empirical approach to its definition and study. BMC Medicine. 2005, 3:19.

[26] Carruthers et al. Myalgic encephalomyelitis: International consensus criteria. Journal of Internal Medicine. 2011 Jul 20. doi: 10.1111/j.1365-2796.2011.02428.x. [Epub ahead of print]

October 13, 2011