Epidemiology and survival of idiopathic pulmonary fibrosis from national data in Canada

Discussion

Based on our analysis of the data from the mandatory national reporting of institutional acute care in Canada, we estimated that there were 14 259 prevalent cases and 6390 incident cases of IPF in the year 2011. This corresponds to a rate per 100 000 population for all ages of 41.8 for prevalence and 18.7 for incident cases. The rate of prevalence was about 19% higher in men (45.4 cases per 100 000) than women (38.2 per 100 000). The rate of incidence was 31% higher in men (21.3 per 100 000) than women (16.2 per 100 000). We report that there are few cases aged <50 years, and rare cases aged <19 years of age may have a genetic origin, despite using similar coding to establish cases.

Previous research has reported the rates of prevalence and incidence in IPF in other jurisdictions using different methods (table S4). A recent literature review pointed out that many of the estimates relied on large extrapolations for the prevalence of IPF [9]. For example, Raghu et al. [6] relied on a health claims database that covered <1% of the population of the USA, while studies conducted based on the UK THIN databases covered 255 general practices, or 3% of the population [21]. Currently, the THIN data contain the electronic medical records of 11.1 million patients collected from 562 general practices in the UK, covering 6.2% of the UK population [22]. If we had relied on previous studies with less data capture, our estimates may have been underreported by 50% or more.

However, there is provincial variation, which suggests that there may also be country level variation in the estimates of IPF prevalence and incidence, if studies similar to our study are conducted in the future elsewhere. The highest rates were seen in the province of Quebec (figure 2), which, along with Ontario, is a major manufacturing industrial province in Canada (populations 13 million and 8 million, respectively). Since the general health of people in the provinces is similar, with life expectancy differences being <2 years between all provinces, variation in the prevalence may be due to differences in the delivery of healthcare services, with the smaller eastern maritime provinces of New Brunswick (0.7 million), Nova Scotia (0.9 million), Prince Edward Island (0.1 million), and Newfoundland and Labrador (0.5 million) perhaps having closer access to hospital-based care instead of dispersed community-based care, creating greater data capture of cases.

There are strengths and limitations in the analysis when using national administrative institutional databases. In our study, we were able to capture 100% of all admissions and day surgeries (including visits for advanced diagnostic testing in hospital). Data coverage was not comprehensive for emergency room visits across all provinces, with only Ontario and Alberta having mandatory reporting. Overall, the additional cases that were identified by extrapolation for missing emergency room visits added 13% for broad prevalence (n=1603) and 21% for broad incidence (n=1119) (figure 1). However, the percentage increases in the number of extra cases that were identified using either Ontario or Alberta data for emergency visits were similar, which strengthened the method of extrapolation of extra cases across Canada based on emergency room visits from these two provinces. Still, since there is provincial variation in healthcare systems, there may be some variability not identifiable in the databases.

The identification of cases was limited by coding of admissions and lack of linkage to physician records. In our analysis, we had access to available ICD-10 CA codes, which may be more specific than ICD-9 codes used in other work in the USA [6]. In particular, the term IPF was not included in the ICD-9 codes, and ICD-9 codes may classify more patients with other common ILDs such as diffuse pulmonary fibrosis. However, the ICD-10 CA code J84.1 is defined as “other interstitial pulmonary diseases with fibrosis”, which includes “diffuse pulmonary fibrosis”, “fibrosing alveolitis (cryptogenic)”, “Hamman–Rich syndrome”, “IPF” and “usual interstitial pneumonia”, with an unknown percentage of diseases other than IPF.

In addition, unlike other work, we did not have access to physician billing records, which are only available in provincial databases. Future research on linking institutional care with physician billings at the provincial level is ongoing, and could produce higher estimates of prevalence and incidence for cases undergoing physician care and not yet requiring acute institutional care, or under physician care without a recent admission. We can speculate on the extra possible cases based on work by Collard et al. [15], where 52.9% of prevalent cases of IPF had an admission and 37.0% had emergency care, leaving the residual cases (10.1%) identified by physician care alone. We are unaware of any epidemiological study that stratifies the case definition by source of care.

Using diagnostic tests to refine the case definition may underestimate the rates of IPF, as the recording of diagnostic tests is not mandatory for admissions, as we have demonstrated for other diseases. For example, in our previous work using national admissions data, we found that there was <1% reported use of radiography for hip fracture, while admissions for hip fracture that require surgical repair often require three radiographs (pre-surgery, post-surgery and follow-up) [23], i.e. diagnostic tests that are common for each type of admission are not always separately reported. Conversely, the administrative databases do not provide test results and the use of CT, biopsy or bronchoscopy may not confirm a J84.1 code and we may have included IPF where the test results may have been negative, which suggests an overestimate. When compared with previous studies, the ratio of narrowly defined cases to broadly defined cases was higher in our study: 48% for prevalence and 48% for incidence. Other studies report 33% for prevalence and 42% for incidence [6, 15]. The key difference may be the level of care that was being provided in our study, with all patients being identified as having any one of an acute care admission, day surgery, emergency room or in-hospital clinic visit. Within these settings, more diagnostic tests may have been ordered. The extent of resource utilisation and economic impact of IPF in Canada warrants further study.

One key finding in our analysis was the high rate of death following the diagnosis of IPF. In 2011, 3051 persons aged ≥50 years died with IPF. For comparison, 10 546 persons aged ≥50 years died of COPD, the fourth leading cause of death, in 2007 [20]. While the rate of death is high for IPF in this analysis, we may have underestimated the risk of death because we only captured hospital/emergency room deaths and may have excluded deaths that occur at home. In contrast, the rate of prevalence of IPF was only 2.9% of the rate of prevalence of COPD. Despite the lower prevalence, the survival from IPF was poor (41% death at 4 years), confirming that the diagnosis of IPF would be devastating and that a delayed diagnosis could be too late to receive any remedial care. Furthermore, the quality of life of IPF patients is already poor and worsens over time.