Idiopathic Pulmonary Fibrosis -- A Systematic Review
Idiopathic Pulmonary Fibrosis -- A Systematic Review
Our systematic search on IPF epidemiological parameters resulted in 13 eligible original articles published after 2001, of which six used the ATS/ERS 2000 diagnostic criteria or classification recommendations of IIP from 2002. Our results indicate that there exist only a few reports on IPF epidemiology according to the modern diagnostic criteria, and that all studies had either major methodological limitations or a small cohort of patients, which makes the comparison of epidemiological parameters in these studies very difficult or impossible. No original articles have been published so far on IPF prevalence according to the new diagnostic criteria (ATS 2011) or from IPF registries. The use of uniform, international diagnostic guidelines is of utmost importance in the field of IPF medicine, and will most likely enable the collection of more accurate and comparable data in the future. It is, however, surprising that quite a few articles have been published after the publication of the first diagnostic criteria in 2000, yet they did not make use of these criteria by means of the reevaluation of patient data. Data reevaluation requires access to a large set of patient data, including histological lung biopsies and original CT scans, and often requires a complete search of all medical records, perhaps making it too laborious and costly. However, as our literature search showed, this has been done in restricted patient cohorts, and could well be standardized as a method for future studies. Uniform electronic patient journals and digital image archives may offer some help in this setting, making the collection and evaluation of epidemiological data easier for scientists in the future. Thus it is important to collect data in such a way that even if the diagnostic criteria change, the data could still be used.
In addition to the identification of epidemiological parameters, our primary aim was to identify plausible and reliable methods in studies related to IPF epidemiology, as the field of IPF medicine is undergoing considerable change due to specific drugs that are entering clinical use, and many nationwide registry databases which are currently being built. Various and very heterogeneous methods are being used by IPF researchers, and so far the data from these studies is not comparable between countries due to this variation. We identified three categories or strategies for collecting IPF epidemiological data: nationwide registries for IPF or ILD, such as the Italian registry; questionnaires for health-care professionals; and systematic searches from existing databases. From these methods, questionnaires seemed to produce the lowest prevalence rates compared to national registries and/or searches from pre-existing databases. The coverage of pre-existing databases might be high in cases such as those shown in the study of Navaratnam, but the downside of this kind of study rationale is the high number of patients that have been misdiagnosed i.e. having a high number of "false positive" cases.
The number of incidental and prevalent cases varied greatly in the presented studies (prevalence from 0.5 to 27,9/100,000). In some of the selected articles e.g., it was speculated that this variation may be due to real differences between countries. Our results suggest that, primarily, the differences in epidemiological parameters are more likely to be a result of the heterogeneous methods used than true geographical differences in IPF epidemiology. The fact that studies which used very similar methodologies resulted in similar results (for example the questionnaire studies, where the incidence ranged from 0.93 to 2.94/100,000) supports this view. However, even these studies are not comparable as the questionnaires used, and the strategies for obtaining answers, were different. The possible differences in results between epidemiological studies are summarized in Table 3. One shared problem in all methods is the difficulty of finding the right denominator to epidemiological calculations. For example in the registry study of Thomeer et al. low prevalence and incidence rates may be due to overly high denominator; not all but 20 centers participated to the study which doubtfully covers the whole Flanders population. In addition, the lack of a specific diagnostic coding for IPF, that would differentiate it from other ILDs has not existed before the 2013 version of the ICD-10 diagnostic system.
Changes in diagnostic criteria may also explain some of the observed variation. Studies implemented before the ATS/ERS 2000 guidelines have shown the lowest prevalence at 0.5 in Taiwan and the highest in Norway at 23.4. After the publication of more definitive diagnostic criteria, prevalence has varied between 3.38 to 27.9. Until the 1990s, the Liebow classification of idiopathic interstitial pneumonia (IIP) was used, which did not include NSIP. NSIP was described in 1994, and subsequently a new proposal of the classification of IIPs, which also included NSIP, was presented by Katzenstein and Myers in 1998. Until this proposal, NSIP was probably diagnosed mostly as IPF or, in some cases, also as desquamative interstitial pneumonia (DIP). The international guidelines of IPF and IIP in 2000 and 2002 recommended that NSIP should be separated from IPF, which has been gradually implemented into the clinical practice during the 2000s. Thus, it is likely that most of the studies in the present review may have also included patients with NSIP in addition to IPF, at least in the beginning of the study periods, because of the historical background and the change in classifications. Currently a new classification of IIPs is being updated which will include a new category, namely unclassifiable IIP, which has been shown to comprise about 10% of an ILD cohort. It may be reasonable to assume that in the above-mentioned epidemiological studies not only the patients with NSIP, but also those with unclassifiable fibrosis, may have been included.
The international disease classifications have also reflected the lack of a clear, systematic classification of ILD:s: only in the latest version of ICD-10 from 2013, there has been a separarate subclassification for IPF. It is to be expected, that studies performed after this disease classification has been taken into use, and in countries where the population registries are most reliable, will yield high-quality data on IPF epidemiology. There are already preliminary reports on several national registry studies, which hopefully will yield comparable data on IPF epidemiology in different geographical and cultural areas.
Discussion
Our systematic search on IPF epidemiological parameters resulted in 13 eligible original articles published after 2001, of which six used the ATS/ERS 2000 diagnostic criteria or classification recommendations of IIP from 2002. Our results indicate that there exist only a few reports on IPF epidemiology according to the modern diagnostic criteria, and that all studies had either major methodological limitations or a small cohort of patients, which makes the comparison of epidemiological parameters in these studies very difficult or impossible. No original articles have been published so far on IPF prevalence according to the new diagnostic criteria (ATS 2011) or from IPF registries. The use of uniform, international diagnostic guidelines is of utmost importance in the field of IPF medicine, and will most likely enable the collection of more accurate and comparable data in the future. It is, however, surprising that quite a few articles have been published after the publication of the first diagnostic criteria in 2000, yet they did not make use of these criteria by means of the reevaluation of patient data. Data reevaluation requires access to a large set of patient data, including histological lung biopsies and original CT scans, and often requires a complete search of all medical records, perhaps making it too laborious and costly. However, as our literature search showed, this has been done in restricted patient cohorts, and could well be standardized as a method for future studies. Uniform electronic patient journals and digital image archives may offer some help in this setting, making the collection and evaluation of epidemiological data easier for scientists in the future. Thus it is important to collect data in such a way that even if the diagnostic criteria change, the data could still be used.
In addition to the identification of epidemiological parameters, our primary aim was to identify plausible and reliable methods in studies related to IPF epidemiology, as the field of IPF medicine is undergoing considerable change due to specific drugs that are entering clinical use, and many nationwide registry databases which are currently being built. Various and very heterogeneous methods are being used by IPF researchers, and so far the data from these studies is not comparable between countries due to this variation. We identified three categories or strategies for collecting IPF epidemiological data: nationwide registries for IPF or ILD, such as the Italian registry; questionnaires for health-care professionals; and systematic searches from existing databases. From these methods, questionnaires seemed to produce the lowest prevalence rates compared to national registries and/or searches from pre-existing databases. The coverage of pre-existing databases might be high in cases such as those shown in the study of Navaratnam, but the downside of this kind of study rationale is the high number of patients that have been misdiagnosed i.e. having a high number of "false positive" cases.
The number of incidental and prevalent cases varied greatly in the presented studies (prevalence from 0.5 to 27,9/100,000). In some of the selected articles e.g., it was speculated that this variation may be due to real differences between countries. Our results suggest that, primarily, the differences in epidemiological parameters are more likely to be a result of the heterogeneous methods used than true geographical differences in IPF epidemiology. The fact that studies which used very similar methodologies resulted in similar results (for example the questionnaire studies, where the incidence ranged from 0.93 to 2.94/100,000) supports this view. However, even these studies are not comparable as the questionnaires used, and the strategies for obtaining answers, were different. The possible differences in results between epidemiological studies are summarized in Table 3. One shared problem in all methods is the difficulty of finding the right denominator to epidemiological calculations. For example in the registry study of Thomeer et al. low prevalence and incidence rates may be due to overly high denominator; not all but 20 centers participated to the study which doubtfully covers the whole Flanders population. In addition, the lack of a specific diagnostic coding for IPF, that would differentiate it from other ILDs has not existed before the 2013 version of the ICD-10 diagnostic system.
Changes in diagnostic criteria may also explain some of the observed variation. Studies implemented before the ATS/ERS 2000 guidelines have shown the lowest prevalence at 0.5 in Taiwan and the highest in Norway at 23.4. After the publication of more definitive diagnostic criteria, prevalence has varied between 3.38 to 27.9. Until the 1990s, the Liebow classification of idiopathic interstitial pneumonia (IIP) was used, which did not include NSIP. NSIP was described in 1994, and subsequently a new proposal of the classification of IIPs, which also included NSIP, was presented by Katzenstein and Myers in 1998. Until this proposal, NSIP was probably diagnosed mostly as IPF or, in some cases, also as desquamative interstitial pneumonia (DIP). The international guidelines of IPF and IIP in 2000 and 2002 recommended that NSIP should be separated from IPF, which has been gradually implemented into the clinical practice during the 2000s. Thus, it is likely that most of the studies in the present review may have also included patients with NSIP in addition to IPF, at least in the beginning of the study periods, because of the historical background and the change in classifications. Currently a new classification of IIPs is being updated which will include a new category, namely unclassifiable IIP, which has been shown to comprise about 10% of an ILD cohort. It may be reasonable to assume that in the above-mentioned epidemiological studies not only the patients with NSIP, but also those with unclassifiable fibrosis, may have been included.
The international disease classifications have also reflected the lack of a clear, systematic classification of ILD:s: only in the latest version of ICD-10 from 2013, there has been a separarate subclassification for IPF. It is to be expected, that studies performed after this disease classification has been taken into use, and in countries where the population registries are most reliable, will yield high-quality data on IPF epidemiology. There are already preliminary reports on several national registry studies, which hopefully will yield comparable data on IPF epidemiology in different geographical and cultural areas.