Background Prospective national screening and surveillance programmes serve a range of public health functions. Objectively determining their adequacy and impact on disease may be problematic for rare disorders. We undertook to assess whether objective measures of disease surveillance intensity could be developed for the rare disorder sporadic Creutzfeldt–Jakob disease (CJD) and whether such measures correlate with disease incidence.
Method From 10 countries with national human prion disease surveillance centres, the annual number of suspected prion disease cases notified to each national unit (n=17 610), referrals for cerebrospinal fluid (CSF) 14-3-3 protein diagnostic testing (n=28 780) and the number of suspect cases undergoing diagnostic neuropathological examination (n=4885) from 1993 to 2006 were collected. Age and survey year adjusted incidence rate ratios with 95% CIs were estimated using Poisson regression models to assess risk factors for sporadic, non-sporadic and all prion disease cases.
Results Age and survey year adjusted analysis showed all three surveillance intensity measures (suspected human prion disease notifications, 14-3-3 protein diagnostic test referrals and neuropathological examinations of suspect cases) significantly predicted the incidence of sporadic CJD, non-sporadic CJD and all prion disease.
Conclusions Routine national surveillance methods adjusted as population rates allow objective determination of surveillance intensity, which correlates positively with reported incidence for human prion disease, especially sporadic CJD, largely independent of national context. The predictive relationship between surveillance intensity and disease incidence should facilitate more rapid delineation of aberrations in disease occurrence and assessment of the adequacy of disease monitoring by national registries.
- Creutzfeldt-Jakob Disease
Statistics from Altmetric.com
National screening and surveillance programmes encompass a broad range of diseases, which collectively serve a diverse spectrum of important public health objectives. Examples of national screening programmes are those aimed at the early detection of cancers or precancerous states, such as in relation to cervix,1 breast2 ,3 and bowel,4 ,5 while infectious or transmissible diseases like HIV,6 hepatitis C7 ,8 and Creutzfeldt–Jakob disease (CJD)9 are a frequent focus of national surveillance centres. Evaluation of national screening and surveillance programmes, using objective measures of the adequacy of screening and the consequent impact on disease incidence, is important for measuring the utility of these public health initiatives.
In relation to national screening for certain malignancies, the ‘at risk’ group can often be defined by age and/or sex, with national census data delineating the size and geographical distribution of the target population. Consequently, the adequacy of screening and impact on disease incidence over time can be determined reasonably easily and accurately, including providing insights into how completeness of disease screening impacts disease incidence.1 Objective assessment of the adequacy and consequent impact of national surveillance is arguably more difficult for rare disorders like sporadic CJD, where screening is not possible and surveillance largely based on the suspicion of disease.
CJD is the most common human phenotype of the rare, transmissible neurodegenerative disorders known as prion diseases. Over the last 15 years, there has been substantial improvement in premortem diagnostic capacity, mainly through CSF 14-3-3 protein detection and brain MRI.10–12 With a reported sensitivity of 85%–90%,11 ,13 the utilisation of the 14-3-3 protein CSF test has greatly increased since its introduction during the 1990s.14
Sporadic CJD incidence is frequently quoted as 1/million population/year,15 ,16 implying a fixed and always maximal incidence, perhaps secondary to intrinsic biological features of the disease, yet officially reported national incidence rates vary from 0.48 to 2.23/million population/year.9 Small populations, more vulnerable to incidence variation, may partly explain such disparities but the question remains as to why countries with inherent similarities in surveillance methodologies and population sizes have varied reported incidence rates. A recent report has suggested that surveillance intensity, measured through assessment of population interrogation rates, correlates with disease incidence.17 This is an intuitively appealing but hitherto unsubstantiated hypothesis, particularly in a multi-national setting.
The primary aim of the current study was to examine within a multi-national context the relationship between reported disease incidence and surveillance intensity, the latter measured through adapting relatively simple monitoring methods into objective metrics.
National surveillance centres constituting the European and Allied Countries Creutzfeldt–Jakob Disease Surveillance Consortium (EUROCJD) were invited to participate in the study. In addition, based on analogous prospective national human prion disease surveillance programmes and similar data collection mechanisms, the Czech Republic and Hungary were also included in the study. Details concerning the EUROCJD participating countries and their ascertainment and data collection methods have been published previously.9 ,11 Overall, 10 countries (Australia, Austria, Canada, Czech Republic, France, Germany, Hungary, Republic of Ireland, Netherlands and the UK) provided data for the 14-year period of 1993–2006, with a key requirement being their ability to undertake detailed age-stratification of the annual national populations for each year of study data supplied. Surveillance activities in the majority of countries involved began in 1993 and the timeframe of data collected was chosen to reflect this fact and the introduction of the CSF 14-3-3 test in 1997, providing a long-term perspective of the study measures assessed. Data requested for analysis were annual counts for each category of: persons referred to national surveillance units with suspected prion disease; those referred with suspected prion disease undergoing neuropathological assessment; and persons referred for CSF 14-3-3 protein testing. For countries commencing national prospective surveillance prior to 1998, CSF 14-3-3 protein detection was available from 1997, while in countries where prospective national surveillance commenced after 1997 (Canada and Czech Republic), data were limited to the available years within the epoch (refer table 1).
Three variables were selected as potential objective predictors of disease incidence: notification of suspected prion disease to a national registry; referrals for diagnostic CSF 14-3-3 protein testing; and referrals for neuropathological examination in suspected prion disease cases. Suspected prion disease notifications included all referrals made to each national surveillance centre between 1/1/1993 and 31/12/2006, regardless of the notification source and the perceived prima facie likelihood of prion disease; the latter was to mitigate potential selection biases. Specific practices for CSF 14-3-3 protein testing varied between countries; some national surveillance centres were notified by multiple delegated diagnostic laboratories of all testing results, while in other countries only positive results were notified by external laboratories and not all samples were tested. In many countries, surveillance centres undertook all testing within their own laboratory and were therefore aware of all referrals, regardless of result. Although this variation in CSF referrals introduces a potential source of bias, it was tolerated to accommodate minor differences in national surveillance formats and to enhance the broader applicability of the study and its findings across the authentic range of national surveillance practices. Referrals for neuropathological examination included all suspected cases of prion disease referred to national surveillance centres undergoing autopsy examination of the brain or brain biopsy, offering a ‘gold standard’ for prion disease diagnosis.
Cases where onset and notification occurred within one country but death occurred elsewhere were excluded from the analysis, as were CSF 14-3-3 protein samples referred from outside the country in question. Within each country, data for the three predictor variables were collected as total numbers and were stratified by survey year into 5-year age groups by: year and age at notification or death (if dead at time of notification) for the suspected prion disease referral variable; year and age at CSF 14-3-3 protein sampling for CSF referrals; and year and age at notification (or death if dead at time of notification) for neuropathological examination of suspected prion disease referrals.
Three outcome variables were determined: the annual number of definite and probable cases of sporadic CJD; non-sporadic CJD (encompassing variant CJD, genetic and iatrogenic prion disease); and all prion disease cases (sporadic plus non-sporadic CJD plus other prion diseases). Human prion disease evinces a heterogeneous phenotype, such that a case may be referred as being suspected variant CJD but ultimately classified as a different form of prion disease. The national surveillance units involved in the study use the same referral systems for all types of prion disease, with final classification of definite and probable cases based on internationally recognised clinical and/or neuropathological criteria. To assess the potential influence of the numerical predominance of sporadic CJD on the all prion disease groups, non-sporadic CJD cases were analysed as a separate group.
After data auditing, the Netherlands was excluded from the modelling analyses due to incomplete data sets for the study period.
Adjustment for age structure differences in the nine countries was performed by direct age-standardisation, with data adjusted to a standard population; the Australian 2000 population was chosen due to it being the median size and midpoint of the data set collected. Crude rates of disease incidence (per million population) were initially calculated for each country. Trends in rates overall and within each country for referrals of CSF 14-3-3 protein detection, suspect CJD cases and neuropathological examinations postmortem were assessed. To investigate the predictors for sporadic, non-sporadic and all prion disease cases, incidence rate ratios (IRRs) were estimated with 95% CIs using random-effects Poisson regression methods to take into account within-country variation for repeated measures across the survey years: reported suspect cases, CSF referrals and performing neuropathology in suspected cases. The test for heterogeneity using the random-effects variance parameter estimate was significant (p<0.01), confirming the appropriateness and suitability of the random-effects model. Stata, V.10.0 (StataCorp, College Station, Texas, USA) statistical software was used in all analyses.
For the period 1993–2006, 17 610 persons with suspected prion disease were notified to national surveillance centres across the nine countries, with 28 780 CSF 14-3-3 protein referrals for diagnostic testing, and 4885 cases of suspected prion disease undergoing neuropathological assessment (table 1). For the same period, there were 4235 (definite and probable) sporadic CJD cases (approximately 84% of the total) among 5054 cases classified with any form of prion disease. Overall, 70% of the 5054 definite and probable CJD cases and 73% of all suspect prion disease referrals who underwent neuropathological examination were classified as definite cases. For the nine countries, national crude annual referral rates of suspected prion disease cases ranged between 0.20 and 26.62 cases per million per year (annual average 1.55–12.32 referrals/million/year), while CSF referrals ranged from 0.20 to 29.41 per million per year (annual average 0.92–17.25 referrals/million/year) and neuropathology examination in suspect cases ranged from 0.19 to 4.84 examinations per million per year (annual average 0.88–2.64 examinations/million/year) (table 2, see online supplementary tables 1 and 2). For sporadic CJD, the crude annual incidence rates varied from 0.19 to 2.29 cases per million per year (annual average 0.77–1.42 cases/million/year) (table 3) and for all prion diseases, the crude annual incidence rates varied from 0.19 to 2.58 cases per million per year (annual average 0.87–1.73 cases/million/year) (see online supplementary table 3).
Temporal trend analysis of each age-adjusted, predictor and outcome variable over the 1993–2006 period was undertaken for the pooled national data. Overall, there was a significant increase in suspect case referrals (IRR 1.09 (95% CI 1.08 to 1.10)), diagnostic CSF 14-3-3 protein detection referrals (IRR 1.18 (95% CI 1.17 to 1.19)), sporadic CJD (IRR 1.03 (95% CI 1.01 to 1.04)) and all prion disease cases (IRR 1.02 (95% CI (1.01 to 1.03)) but no increase in neuropathological examinations of suspect cases at postmortem (IRR 1.00 (95% CI 0.99 to 1.01)) over time (see online supplementary table 4). Differences between countries in terms of the temporal trends and the correlations between surveillance intensity predictor variables and disease incidence were observed over the study period. Seven of the nine countries displayed congruence with the principle that significant increases in the average annual trend of suspect prion disease notifications combined with diagnostic CSF 14-3-3 protein detection referrals correspond with increases in disease incidence (sporadic CJD and/or all prion disease) over the study period. Increasing surveillance variables in Hungary was associated with an increase in only the ‘all prion disease’ category, and in one country (Czech Republic) the conformity to this principle was that no increase in surveillance intensity variables over the study period corresponded with a lack of observed increase in disease incidence. In only two countries (Australia and Republic of Ireland) was there no correspondence between a significant increase in the average annual trend in suspect prion disease notifications and diagnostic CSF 14-3-3 protein detection referrals over the study period and increases in either sporadic CJD and/or all prion disease incidence; this lack of correlation did not clearly correspond with the magnitude of the IRRs for surveillance intensity.
The test for heterogeneity using the random-effects variance parameter was significant (p<0.001), indicating suitability for employing random-effects Poisson modelling. After age adjustment, the analysis showed highly significant correlations between each predictor variable and the incidence of disease (sporadic CJD, non-sporadic CJD and all prion disease groups) (table 4), with the association between the neuropathological examinations of suspect cases as predictor and disease incidence less substantial than the other two predictor variables.
By age and survey year adjusted analysis, the associations between the predictor and outcome variables (table 4) showed suspect case notifications, diagnostic CSF 14-3-3 protein detection referrals and neuropathological examination each independently predicted sporadic CJD, non-sporadic CJD and all prion disease incidence. The only exception was a modest negative association between diagnostic CSF 14-3-3 protein detection referrals and non-sporadic CJD. In comparison with the level of significance of our other analyses we believe this modest result may imply a lack of meaningful correlation. We suspect the outcomes observed for the all prion disease group most likely represent a carryover effect through the numerical predominance of sporadic CJD in this group.
This is the first study to directly address and verify previous suggestive findings17 and prevailing assumptions that greater scrutiny of a population may lead to higher rates of human prion disease detection and thereby reported disease incidence. Integral to the more objective assessment approach described in our study is the adjustment of surveillance detection methods as population interrogation rates. Employing this method, we have shown that relatively simple, accessible measures (clinical case recognition with notification made to a national surveillance centre combined with a routine diagnostic test like CSF 14-3-3 protein detection and neuropathological examination) can be adjusted to generate metrics of surveillance intensity, which correlate sufficiently with sporadic CJD incidence to be predictive. The broad multi-national context, very large number of cases included and the extended time period of the study all add support for the robustness of the findings. As such, our observations provide important insight into factors that contribute to variations in reported prion disease incidence, particularly for sporadic CJD.9 Nevertheless, in the absence of universal autopsy, some uncertainty inevitably persists regarding the ascertainment of all human prion disease cases and the absolute incidence of CJD.
The observed predictive relationship between surveillance intensity and sporadic CJD incidence was present despite the recognised clinical profile variations across the sporadic CJD subtypes,11 ,18 suggesting that national surveillance can be achieved for a disease through primarily relying on clinical detection even when there is a somewhat diverse phenotypic spectrum. It is also of interest that advanced age, well beyond that more typically associated with sporadic CJD, does not appear to impede clinical recognition, with previous studies suggesting that increased incidence appears to correlate with increased referrals of suspect cases in the very elderly.17 ,19
The epidemiological history of human prion diseases, especially acquired forms such as those related to the therapeutic use of cadaveric dura mater explants and pituitary hormones20 as well as variant CJD21 and ongoing concerns regarding transmission risks posed by routine surgery, including non-neurosurgery,22 ,23 should serve as a sobering precaution against any public health complacency in relation to this group of diseases. The key objectives of prospective national surveillance programmes are to ensure adequate detection of the disease and accurate depiction of the epidemiological profile so that significant departures in disease incidence or demographic characteristics can be recognised.24 Our findings should subserve both of these objectives. As described for variant CJD,21 an altered phenotype can indicate important epidemiological changes with respect to disease aetiology. Notwithstanding this influence, a more accurately defined and predictive relationship between population surveillance intensity and disease recognition rates (with 95% CIs) should facilitate more rapid and confident delineation of significant deviations in national disease incidence for a given surveillance intensity and serve to prompt more detailed evaluation of why this change has occurred, possibly leading to more expeditious deployment of public health responses.
Our study revealed major variations, approximately two orders of magnitude, in annual crude notification rates of persons with suspected prion disease across the study period in the countries assessed. Although some of this variation may be a natural finding based on variable population sizes, it demonstrates the likely underappreciation of true differences when only drawing comparisons of unadjusted, average disease incidence rates from countries with varied population sizes. Over the study timeframe, CJD awareness and recognition increased markedly both nationally and globally due to factors including the bovine spongiform encephalopathy (BSE) epidemic, improved surveillance systems and advances in diagnostic capabilities, in particular the CSF 14-3-3 diagnostic test. These influences have contributed to improved notification and diagnostic assessment of suspected cases, a feature reflected in the positive temporal trends observed for suspect case and 14-3-3 CSF test referrals and, to a lesser degree, with neuropathological examination in the participant countries. Although temporal variation in predictor and outcome variables was significant between the countries over the study period, the analysis of pooled data, adjusted for time and age, demonstrated that time was independent of the association between surveillance intensity measurements and incidence, further underscoring the strength of this study.
Referrals for 14-3-3 CSF analysis in France and Germany were noticeably higher than in other countries. We surmise this may be due to subsidisation of the costs of testing and/or the strong tradition of employing CSF surrogate biomarker detection in the evaluation of dementias, including CJD. In Hungary and the Czech Republic, the lower number of referrals for 14-3-3 testing is due to the later point at which 14-3-3 testing was introduced and adopted into the criteria in comparison with France and Germany. However, despite the higher levels of CSF testing in Germany and France, incidence was not increased in comparison with other countries, suggesting that the correlation with sporadic CJD incidence also relies on referral of suspect cases to a national reference centre, as well as neuropathological examination. Furthermore, this finding argues against a potential misconception from this study that widespread 14-3-3 CSF testing is required for optimal case ascertainment. The negative association observed between CSF referrals and non-sporadic CJD cases may relate in part to modest increases in prion protein gene (PRNP) testing observed over the study timeframe, translating into a reduced likelihood of non-sporadic CJD cases (mainly genetic) undergoing 14-3-3 CSF testing. This is a plausible explanation, however, unverified in this study, with the negative association also possibly related to the fact that CSF 14-3-3 protein detection was primarily developed for sporadic CJD detection.
The multi-national context of the present study was an intended and important design feature by which it was ensured relatively minor but ‘real life’ variations in specific mechanisms of national prion disease surveillance were encompassed. Despite this, the findings clearly demonstrate overall support for a positive and predictive correlation between surveillance intensity and sporadic CJD incidence; however, we note that this predictive association was not observed in two of the nine countries. The reason for this is uncertain but may relate to unapparent variations in surveillance methods employed within these countries or a potential saturation point where no new cases are identified despite increasing surveillance intensity.
Although the public health and epidemiological context of each disease undergoing national screening or surveillance is relatively unique, and therefore of limited direct comparability, the finding that observed population diagnosis rates can be influenced by testing and resourcing issues is not unique to CJD. HIV infection diagnosis in the UK is estimated at 74% of all infections, while in Australia, the estimate is 85%–90%.25 ,26 The Australian epidemic remains largely transmitted through male homosexual contact, a well-informed group with a high rate of HIV testing.27 In contrast, the epidemic in the UK is more heterogenous and often associated with migration to the UK from high prevalence countries.28
So, in conclusion, our study has confirmed that routine national surveillance methods adjusted as population rates allow objective determination of surveillance intensity, which correlates positively with reported incidence for human prion disease, especially sporadic CJD, largely independent of national context. The predictive relationship between surveillance intensity and disease incidence should facilitate more rapid delineation of aberrations in disease occurrence and permit objective assessment of the adequacy of disease monitoring by national registries. Further, the approach outlined in our study could extend to other rare disorders, and should allow more objective assessment of the adequacy of population scrutiny by extant or potential members of surveillance consortia.
Australia: The Australian National Creutzfeldt-Jakob Disease Registry (ANCJDR) wishes to thank the families and managing physicians of patients with prion disease for their generous support. Austria: The Austrian Reference Center for Human Prion Diseases (ÖRPE) gratefully acknowledges the continuous support by Austrian neurologists, neuropathologists, other clinical personnel and families, without whom CJD surveillance would be impossible. Czech Republic: National Reference Laboratory for Diagnostics of Human Prion Diseases would like to thank the families of patients with prion disease and involved physicians for their help and support. France: The French National Surveillance Network for CJD wishes to thank all physicians for cases notification, biochemical laboratories for 14-3-3 protein detection and neuropathologists for pathological data. Hungary: The Hungarian Prion Disease Reference Center wishes to thank the families, the managing physicians of patients with prion disease and the National Center for Epidemiology (Dr Zsuzsanna Molnar) for their generous support. Ireland: The staff of Ireland's CJD Unit wish to acknowledge the help and cooperation of the family members of patients diagnosed with CJD in Ireland over the last 20 years. Netherlands: We thank all patients, families and neurologists for their participation, and our colleagues at the National Surveillance Center for Prion Disease, part of the Department of Pathology at Utrecht University Medical Center, for their close collaboration in obtaining autopsy results; and the contributions by M.Schuur to this study were greatly appreciated. UK: The National CJD Research and Surveillancex Unit is funded by the Policy Research Programme in the Department of Health. The views expressed in this article are not necessarily those of the Department.
Review history and Supplementary material
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Contributors SC and GK conceived the study, coordinated the data collection, analysed the data and cowrote the manuscript. MS, AB and CLM assisted with draft preparation and editing of manuscript. HW and ML performed statistical analysis and draft preparation. RM, RH, MF, MB, IZ, JMc, RW, J-PB, AA, HB, GGK, GHJ and MC collected, organised and collated case data for analysis and reviewed the manuscript. CI-V and CV reviewed and edited the manuscript. As the principal author, Professor Steven J Collins takes full responsibility for the data, analyses, interpretation and conduct of the research. Professor Collins has full access to the analysed data and has the right to publish these data.
Competing interests None.
Funding GMJAK, MS, AB, CLM: Funded by the Australian Commonwealth Department of Health and Ageing. HW, ML: The Kirby Institute is funded by the Australian Government, Department of Health and Ageing. MS: Funded by the Commonwealth Department of Health and Ageing. RM: Funded in part by IGA NT/14145/2013 grant from the Czech Ministry of Health. RH, MF: The National CJD Surveillance Unit in Ireland is funded by the Department of Health & Children. MB, IZ: Supported by grants from the Robert Koch-Institute through funds of the Federal Ministry of Health (grant no 1369-341), the European Commission (PRIORITY KBBE-2007-2-4-06 ‘Protecting the food chain from prions: shaping European priorities through basic and applied research’, Project number: FP7-KBBE-2007-2A) and by the Verein zur Förderung der Wissenschaft und Forschung an der Neurologischen Klinik Göttingen e.V. CvD, CI-V: The Dutch National Prion Disease Registry is funded by a grant from the National Institute of Public Health and the Environment, an institute of the Ministry of Health, Welfare and Sport. JMcK, RGW: The National CJD Research and Surveillance Unit is funded by the Policy Research Programme in the Department of Health. JPB, AA: The French National Surveillance Network is funded by the Institut National de Veille Sanitaire. HB, GGK: The Austrian Reference Center for Human Prion Diseases (ÖRPE) is funded by the Austrian Federal Ministry for Health. GHJ, MC: Funding of surveillance, laboratory reference services and public health for human prion diseases in Canada is provided through the Prion Diseases Program of the Public Health Agency of Canada. SJC: Funded by the Australian Commonwealth Department of Health and Ageing and is supported in part by an NH&MRC Practitioner Fellowship #APP1005816.
Ethics approval The Australian National CJD Registry has approval for publishing de-identified epidemiological studies relating to prion disease (The University of Melbourne Human Research Ethics Committee # 941450), and as a coordinating institution for the study takes primary ethical responsibility. Other collaborating countries have given similar ethical approval.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.