FormalPara Key Points

Acquiring knowledge on the overall burden of idiopathic pulmonary fibrosis (IPF) is essential for stakeholders planning resource allocation across many conditions. This study provides an overview of the evidence on health-related quality of life (HRQoL) and costs in IPF.

Several studies showed that IPF has a considerable impact on patients’ HRQoL, including physical and social components, in comparison with the general population.

Compared with the national health expenditure or control-matched patient cohorts, IPF was associated with an excess healthcare cost.

Our findings confirm IPF as a growing threat for public health worldwide, with considerable impact to the patients and healthcare providers.

1 Introduction

Idiopathic pulmonary fibrosis (IPF) is a specific form of chronic, progressive fibrosing interstitial pneumonia of unknown aetiology associated with significant morbidity and poor survival [1]. The symptoms include dyspnoea, dry cough, tiredness, aching of muscles and joints, unintended weight loss and finger clubbing [1]. The progression of the disease varies significantly between patients and depends on many clinical and external factors [2]. Overall, individuals with IPF have similar life expectancy to those with non-small cell lung cancer, with reported estimates of median survival being 50% at 3 years and 20% at 5 years post-diagnosis [1, 3,4,5]. The estimates of incidence and prevalence of IPF vary depending on the definition used, the study design, and the underlying population characteristics (such as age, gender, geographic location, etc.) [3, 6]. In general, studies agree that the condition is more common in men and in older people. In Europe, the British Thoracic Society estimates that the prevalence is around 50 per 100,000 population, with the highest rates in Northern Ireland, North West England, Scotland and Wales [7]. This is considerably higher than older estimates from other parts of Europe such as Norway (19.7–23.9/100,000) [8] and Belgium (1.25/100,000) [9]. In North America, two US studies placed the prevalence estimates between 42.7 [10] and 63 [11] patients per 100,000 population (using the broad definition); while a more recent Canadian study reported the prevalence to be as high as 115/100,000 (broad definition) [12]. Similarly, in Japan studies suggested prevalence estimates from 2.9/100,000 in 2005 [13] to 10/100,000 population in 2007 [4]. It follows that, although IPF is still treated as a rare condition in many countries, the evolution of diagnostic methods and greater physician awareness around the disease and an aging population may be leading to an increase in the prevalence and incidence rates over time [6, 14, 15].

There is also considerable activity in the development of treatments for the condition. Before 2010 there was no licensed pharmacological treatment for this devastating disease [1]. In 2008, pirfenidone was approved in Japan and in 2011 by the European Medicines Agency (EMA). In 2014 the US Food and Drug Administration (FDA) approved both pirfenidone and nintedanib, with EMA also confirming approval for nintedanib soon after [16,17,18].Footnote 1 Despite the recent termination of the clinical trial programmes for tralokinumab [19] and simtuzumab [20], a number of new agents are being tested in experimental trials for the treatment of IPF (SAR156597 [21], lebrikizumab [22], FG-3019 [23], PRM-151 [24] and others).

For healthcare providers, who often have to make difficult decisions about resource allocation across many conditions, in-depth knowledge of the overall burden of the disease is essential. Our study involves a comprehensive review of the literature for evidence on health-related quality of life (HRQoL) and costs. It also attempts a qualitative comparison with estimates of HRQoL for the general population and national healthcare expenditure to illustrate the burden of illness of IPF.

2 Methods

The study followed the PRISMA (Preferred Reporting Items for Systematic review and Meta-Analysis) guidelines.

2.1 Search Strategy

Two separate systematic reviews were conducted for economic evaluations and HRQoL evidence. Using the Ovid interface, the databases EMBASE, MEDLINE and MEDLINE In Process were searched for relevant studies. Search terms included disease-specific, economic or cost, and HRQoL keywords such as ‘idiopathic AND pulmonary AND fibrosis’, ‘fibrosing alveolitis’, ‘interstitial pneumonia’, ‘costs and cost analysis’ and ‘health care costs’, ‘HRQoL’, ‘EQ-5D’.Footnote 2

A review of HRQoL was conducted in August 2014 for the development of an economic analysis [25]. All the relevant records from the 2014 review were retrieved and the searches were updated from January 2014 to April 2017.

The economic data search was conducted from database origins to April 2017.

All references were imported into Endnote and duplicate citations were removed.

2.2 Study Selection

A review protocol with inclusion and exclusion criteria was developed at the outset of the study. The inclusion criteria were for adult patients with IPF without any restrictions on the therapy received. Other criteria included the reporting of unit costs, resource use, and HRQoL measures. To increase homogeneity in the study population characteristics, we excluded records that reported costs of diagnosis of interstitial lung disease (ILD).

The protocol was modified during the study to exclude abstract-only records published before 2015 (most often conference proceedings). Those records rarely provided sufficient information on methods and results that could be useful in our study and in general lack the scrutiny of full journal articles. Nevertheless, more recent records (post-2014) were included in our study, as we assumed that at the time of our search they were in development to a manuscript.

Screening of records was conducted in two phases (title/abstract and full-text). One experienced reviewer covered each dataset of records for economic evaluations and HRQoL evidence (EW and KV, respectively). A quarter of the records were screened independently by a second reviewer (AD, LC). If the decision for inclusion or exclusion was different in more than 10%, the full set of records were reviewed again. Because of a > 10% disagreement in the HRQoL dataset, all records were screened in a double-blind manner. The bibliography of another literature review study [26] was used to validate our findings.

2.3 Data Extraction and Analysis

Key pieces of information from the selected studies were extracted in piloted tables by three experienced researchers (EW, KV, LC). A quality check of the data extraction was done by AD. The tables were different for HRQoL and economic evidence. Given the heterogeneity of the economic evidence, we later separated studies that reported healthcare resource use or costs from economic evaluations (cost-effectiveness or budget impact analyses).

3 Results

The database searches identified a total of 3241 records. After removing duplicate records, 2496 abstracts were screened against the eligibility criteria. Twelve additional records were identified via bibliography searches.

A total of 127 publications were included in the qualitative analysis, referring to 66 HRQoL and 28 economic studies. The economic studies were further categorised, with 18 reporting resource use or costs and 10 reporting on cost-effectiveness or budget impact analyses. The overall breakdown of the screening process in the reviews is presented in a PRISMA flow diagram (Fig. 1).

Fig. 1
figure 1

PRISMA flowchart. HRQL health-related quality of life, HCRU healthcare resource use

Full size image

The studies on HRQoL increased over time with almost half conducted and published in the 3.5 years between 2014 and 2017 (see Fig. 2).Footnote 3 We did not identify any cost or economic evaluation studies conducted before 2010, while more than half of the cost studies were published in the last 3 years.

Fig. 2
figure 2

Summary of studies by publication date. HRQL health-related quality of life

Full size image

In terms of geographic regions, the majority of the studies were conducted in Europe and North America (USA and Canada) (Fig. 3). The most studied country was the USA with 13 HRQoL [27,28,29,30,31,32,33,34,35,36,37,38,39,40] and eight economic evidence publications [41,42,43,44,45,46,47,48]. From low income and lower middle income countries (using the World Bank definition [49]) we identified two studies on HRQoL from Egypt [50, 51] and one from India [52]. From east Asia the predominant country was Japan with nine HRQoL studies [53,54,55,56,57,58,59,60,61]; one study was identified from China (HRQoL) [62] and one from Korea (costs) [63]. In the HRQoL dataset, for a number of studies we did not identify a clear country of origin [64,65,66,67].

3.1 Health-Related Quality of Life Evidence

A total of 66 studies were identified (33 in the pre-2014 analysis and 33 post-2014) with HRQoL data in IPF populations. Details of the study location, the population, the HRQoL assessment tools used, and the time points, as well as the sources of funding, are presented in Table 1.

Table 1 Summary of HRQoL evidence
Full size table

In all studies, apart from Jastrzebski et al. [69], the population mean age was over 50 years old, with the average age around 65–70 years old. The study populations were predominantly male with the exception of three studies reporting a higher proportion of female [32, 51] or an equal male/female ratio [30].

The majority of the studies used the disease-specific HRQoL instrument, St. George’s Respiratory Questionnaire (SGRQ), reported in 41 studies. Most of the studies measuring HRQoL with the SGRQ reported results for the three categories: symptoms, impact and activity; in addition to the total score. Despite the development and validation of an IPF-specific version of the SGRQ, the SGRQ-I [70], most investigators, apart from Gaunaurd et al. [28, 71, 72], continue to use the original version.

In addition, six studies reported other disease-specific HRQoL scores such as A Tool to Assess Quality of life in IPF (ATAQ-IPF) [37] or the King’s Brief Interstitial Lung Disease (K-BILD) [73]. The 36-Item Short Form Survey (SF-36) was reported in 26 studies, the EuroQol 5-level questionnaire (EQ-5D) in four studies [39, 40, 67, 74, 75], the SF-12 in two studies and one Canadian study reported Health Utilities Index Mark 2 (HUI2) scores. One study was assessing the mapping of SGRQ data to EQ-5D [76] and another study provided a mapping algorithm from SGRQ data to SF-36 [77]. Further, EQ-5D estimates from phase III trials with nintedanib in IPF (INPULSIS® I and II) were available from an economic evaluation identified during the economic data search [25].

Table 2 reports on a subsection of the studies we found that included HRQoL values based on multi-attribute preference-based measures (EQ-5D and HUI2). We obtained population reference scores for EQ VAS and EQ-5D from a survey conducted across 24 countries [78]. The survey presented scores by age and we selected the 65–74-year age category as the most representative of the IPF studies that we are using in our comparison. To obtain a reference for HUI2 scores, we looked at the US National Health Measurement Study (NHMS) using the scores for ages 65–74 years [79].

Table 2 HRQoL burden of IPF
Full size table

Overall, the HRQoL was found to be lower for patients with IPF compared with the general population (Fig. 4). In the German registry, INSIGHTS-IPF, the EQ VAS of the patients with IPF, was about 9 points lower on the scale compared with the population reference data [80,81,82,83,84]. The difference in the EQ-5D index score was 0.223 lower than the reference. The incremental difference between patients with IPF and the population reference is smaller in the US study STEP-IPF: around 7.5 points on EQ VAS and around 0.1 on EQ-5D index scores [67]. Furthermore, in the study by Rinciog et al. [25], the reported difference in EQ-5D index score ranges from a category with relatively good lung function (forced vital capacity [FVC] > 90% predicted: 0.84) to very poor (FVC < 50% predicted: 0.67).

Fig. 3
figure 3

Regional distribution of identified studies. HRQL health-related quality of life. Asterisk indicates the location was not clearly reported in the study

Full size image
Fig. 4
figure 4

EuroQol 5-level questionnaire (EQ-5D) in patients with idiopathic pulmonary fibrosis (IPF) compared with the general population (reference). FVC forced vital capacity. Asterisk indicates data from by Rinciog et al. were available by FVC% predicted status. The lowest and highest of the available intervals are shown in the figure [25]

Full size image

On the HUI2 instrument, the IPF population utility estimates were substantially lower than those measured on the EQ-5D scale, both for the first year with IPF (0.585) and the fourth year (0.432) [12]. However, some of the difference with the reference scores may be attributed to country variations (US data were used for HUI2 reference).

Regarding other multi-attribute instruments, eight studies reported the average score or the mental and physical component scores (MCS and PCS) of SF-36 [27, 29, 34, 35, 39, 40, 67, 69, 85, 86]. One study reported an SF-36 score of 32 ± 11.4 for severe IPF (defined as diffusing capacity of the lungs for carbon monoxide [DLCO] < 30%) and 59.1 ± 17.8 for patients with mild-to-moderate IPF (DLCO > 30%) [27]. King et al. reported the SF-36 score of 45.7 for placebo and 45.2 for people treated with bonsentan [86]. At baseline, SF-36 PCS scores varied between 26.0 ± 8.0 [85] to 40.6 ± 9.3 [40], with an average value of 35 and SF-36 MCS ranging from 42 [69] to 55.7 ± 7.4 [40] with an average value of 48. The 17 remaining studies detailed the SF-36 results by questionnaire items (physical functioning, social functioning, mental health, role limitations due to physical problems, role limitations due to emotional problems, vitality, bodily pain, and general health perceptions).

3.2 Cost and Healthcare Resource Use Evidence

A total of 18 studies were identified with HCRU and cost evidence (Table 3). The majority were retrospective cohort analyses of claims data. Three studies were based on a synthesis of HCRU and national costs or tariffs [87,88,89]. One study was based on randomised clinical trial evidence [90] and one study was based on clinical expert opinion [91].

Table 3 Summary of cost and resource use studies
Full size table

The most common reported resource or cost was hospitalisation (all-cause and/or respiratory-related), emergency room visits, and acute IPF exacerbation events. The majority of the studies [14] reported costs alongside resource use. Four studies reported only HCRU data [47, 90, 92, 93].

Eight of the studies that reported costs presented estimated total cost per capita [41, 42, 44, 63, 88, 91, 94, 95] (see Table 4). In three US studies the annual total cost of IPF was estimated at around US$20,000 per patient [41, 42, 44]. Controlled for the year the studies were conducted, this estimate was around three times the national per capita health expenditure [96].

Table 4 Cost burden of IPF
Full size table

In 2012, Collard et al. [42] also presented the total costs per person-year for patients with IPF and a matched control cohort (US$26,378 vs US$14,254). In a different study, published a few years later (2015), Collard et al. [41] showed similar estimates of the difference between patients with IPF and controls (US$20,887 vs US$8932).

In a study from Canada [94], the annual cost per patient with IPF was lower than the US studies [41, 42, 44]. However, in relative terms the study estimated a > 3 times greater cost when comparing with the per capita Canadian national heath expenditure.

The annual total cost per patient in Korea [63] was estimated to be < 10% of the cost presented in the American studies [41, 42, 44]. In the same study, the contribution of hospital admission costs to the total healthcare cost was found to be 86.7–88.8%. We also found great disparity in the estimates of the two studies from Spain [91, 97].

An abstract by Hill et al. [88] conducted a bottom-up cost analysis of service provision costs (excluding treatments) in England (NHS) in 2014. They estimated that the actual cost of services was over 40% of the tariff reimbursed by the NHS for each patient with IPF.

From the studies that reported resource use, Wu et al. [93] presented evidence of HCRU in US patients with IPF compared with a matching control cohort (1:3 matching ratio). They found that the mean differences between patients with IPF and control were more pronounced in outpatient hospital visits (7.5 vs 2.7), physician office visits (16 vs 7.8), and oxygen therapies (7.8 vs 0.6). After a multivariate adjustment, the magnitude of the difference was reduced for the outpatient hospital, physician and emergency room visit statistics. Nevertheless, it remained significantly higher for patients with IPF versus non-IPF.

Only five studies reported treatment costs [42, 44, 63, 89, 91]. In Kim et al., treatment costs were between 8–10% of the total costs [63]. However, it was not reported which treatment was considered. In the remaining studies, treatments included corticosteroids, oxygen therapy, azathioprine, cyclophosphamide, N-acetylcysteine (NAC), pulmonary rehabilitation therapy and lung transplantation. Of the new treatments, in Morell et al. it was reported that pirfenidone was offered to patients with IPF on compassionate grounds; it is unclear whether the cost of pirfenidone contributed to the treatment costs in that study [91].

3.3 Economic Evaluations

Ten studies were identified assessing the cost effectiveness, or budget impact, of specific treatment interventions. Details of the methods and results of the studies are presented in Table 5. Three studies were from the UK [25, 26, 98], while the remaining were from France [99], Greece [100], Italy [101, 102], Spain [103], Mexico [104] and USA [105]. The comparators included triple therapy (azathioprine, NAC and steroids), a combination of triple therapy and a genotypic assay thiopurine S-methyltransferase (TPMT), co-trimoxazole, sildenafil, pirfenidone, nintedanib and best supportive care. Only one economic evaluation included lung transplantation as an option for patients [26].

Table 5 Summary of economic evaluations
Full size table

Most studies used a model to synthesise clinical, HRQoL and cost evidence. Moreover, the majority of the analyses used the direct healthcare perspective. Wilson et al. [98] conducted an economic evaluation alongside a multi-centre, randomised, placebo-controlled, double-blind trial of 12 months duration, and reported cost-effectiveness results on both the healthcare direct medical and societal perspectives.

In the economic models, the time horizon ranged between 1, 5 and 30 years, and patient lifetime. A state transition model was used for all papers, and when reported, results were calculated by a cohort analysis. In the long time-horizon models, the cost results varied between US$4000 (£3000) for BSC, US$7000 for NAC and over US$90,000 for new treatments such as pirfenidone and nintedanib. HRQoL benefits ranged between 3 and 4 QALYs. There was a noticeable distinction in the cost effectiveness of old pharmacologic technologies such as triple therapy or NAC, with estimates between US$5000–US$70,000 per QALY, and that of new treatments that exceeded US$100,000 per QALY.

4 Discussion

This was a review of HRQoL, resource use, costs and treatment cost-effectiveness studies conducted over the last 20 years in many countries, and with a variety of objectives, sources of data, and methodologies. As such, it is difficult to express with one coherent estimate the burden of illness of IPF. Nevertheless, several trends appeared in both quality of life and costs.

As with other respiratory conditions, the impact of IPF is not only limited to a worsening of the patient’s breathing function. It has wider consequences for HRQoL including physical (body weight loss, fatigue, clubbing) and social ones (recreational activities, relationships etc.). When reviewing the HRQoL evidence, this review reported on most instruments used in the literature, but focused on generic preference-based measures (such as EQ-5D) to quantify the burden of the disease. By using EQ-5D it is possible to make a comparison between the HRQoL levels with the condition versus the general population, and a comparison across other non-respiratory diseases. Furthermore, EQ-5D is increasingly used in health economic evaluations to calculate quality-adjusted life-years (QALYs), and this work presents a comprehensive review of the available evidence.

Despite the regional differences, there was some agreement between study estimates on the absolute level of HRQoL for patients with IPF; in EQ-5D, scores varied between 0.67 (± 0.242) [67] and 0.8 (± 0.2) [106]. To put this in context, the EQ-5D of patients with arthritis/rheumatism/fibrositis was reported to be 0.597 (CI 0.584–0.609; N = 4145), with hypertension/high blood pressure 0.777 (CI 0.765–0.788; N = 3172) and with asthma 0.797 (CI 0.779–0.814; N = 2452) [107, 108]. In the studies analysed, the decrement in HRQoL for patients with IPF compared with the reference population statistics was between 0.1 and 0.2 points in the EQ-5D.

With regards to costs, three US studies produced comparable estimates of costs per patient around US$20,000 [41, 42, 44]. After adjustments for the study years and currency, the suggested annual per capita cost of IPF patients in North America was estimated between 2.5–3.5 times the national health care expenditure.

We observed discrepancy in the estimates coming from two Spanish studies. This is probably attributed to the methods used. Pedraza-Serrano et al. [97] used data from a Spanish National Hospital Database (CMBD, Conjunto Mínimo Básico de Datos) and conducted a retrospective, descriptive, epidemiological study. Morell et al. [91] took a different approach by synthesising expert opinion from 15 clinicians with unit costs from national formularies. Moreover, Morell et al. [91] included treatments costs, although treatment allocation was not reported. The two estimates are very different to values from the other countries (in absolute and relative terms), which makes it very challenging to select the most accurate. The study by Pedraza-Serrano et al. [97] follows the general trend of a higher per annum cost than the national health expenditure.

Among the cost evidence identified in the literature, we emphasised the existence of matched control cohort studies [41, 42, 93]. These papers provided a direct comparison of the excess costs and resource use of IPF patients versus a reference population. Given that these studies were large in sample size and from a contemporary (2012 and 2015) and generalisable database, they produced relevant estimates for the cost burden of illness of IPF. Therefore, we recommend the use of control or reference cohorts when conducting cost analyses as it provides the relevant benchmark for comparison with the general population.

Two studies also suggested a strong correlation between acute exacerbations of IPF and other external conditions such as seasonality. Collard et al. [41] reported that acute exacerbations of IPF become more frequent in spring and winter. Kim et al. [63] highlighted spring as the season with most events, and linked that to the yellow dust phenomena occurring during that period in Korea, where this study was conducted.

The reader should note the relevance of national guidelines and prescription rules when comparing costs from different countries. Countries with a single (public) payer system, like the UK, have different practices and prescription rules to multiple-payer systems such as Germany in Europe or the US. It is also relevant to consider that some countries may have delayed access to new treatments; for instance, Australia only gained access to new anti-fibrotic agents in 2017, while Europe and the US has had access since 2010–2015.

The evidence on treatment economic evaluations was sparser. The cross-comparison of cost-effectiveness analyses is often hindered by different methodologies, time horizons, approaches in the presentation of the results and many other factors. On this occasion, an additional challenge was that most studies were published only as conference abstracts and, as such, provided little information on their methods and results. This made any comparison or synthesis of cost-effectiveness estimates very difficult.

One omission of our cost estimates is related to the diagnosis of IPF. The diagnostic procedures are largely in common with other ILDs and in most diagnostic cost studies evidence was presented from a heterogenous cohort that included patients with IPF as a subgroup [87, 109]. To include only studies that had an IPF subgroup may have been a misrepresentation of the actual management costs. For internal consistency with our population criteria, we decided to keep the reference database specific to IPF and excluded diagnostic cost studies from our review.

Our qualitative comparison of HRQoL and cost estimates with population reference statistics has further limitations. The synthesis of evidence from various studies involved the comparison of different EQ-5D versions (3L vs 5L) and conversions of cost estimates to one currency. This required several assumptions about the comparability of the data.

This review excluded relevant conference proceedings (published only as abstracts) before 2015. Records published since 2015 were included. Although the information from an abstract is often limited and the research lacks the scrutiny of an academic journal, we considered it important to include more recent records that report relevant information and that could later be published as full manuscripts. This improves the comprehensiveness of the records presented in this review.

However, the inclusion of abstracts could bias the synthesised data used to estimate the burden of illness. For instance, in the HRQoL studies we included data from the INSIGHTS-IPF registry [80,81,82,83,84] and Fell et al. [12] that at the time were available only as abstracts. In the cost studies we included Hill et al. [88] and Mittmann et al. [94].

In our search for evidence on the burden of IPF, we identified other similar literature reviews. Loveman et al. [26] conducted a systematic review with the objective being the comparison of the clinical effectiveness and cost effectiveness of IPF treatment interventions. Our study was not searching specifically for treatment effects, although there was a lot of overlap in our searches for HRQoL and economic evaluations; we identified the same papers in HRQoL and economic evaluations as Loveman et al. In addition, we have used Loveman et al. to validate our review findings [26] within the overlapping time periods.Footnote 4

Lee et al. [6] reported on the unmet public health need with IPF. Although they cover quality of life and resource utilisation, their analysis on the burden of the disease was focused more around the epidemiology, comorbidities and symptoms of IPF.

The treatment of IPF has changed substantially in recent years, and has evolved a lot since the first paper identified in our search was published (2000). We identified an exponential growth of publications in the last 3–5 years. This trend probably follows the development of new pharmacological interventions such as pirfenidone and nintedanib. For instance, we identified many publications referring to results from three nintedanib clinical trials—TOMORROW, INPULSIS® I and II [25, 110,111,112,113,114,115,116,117,118,119,120,121].

With the exception of the evidence reported in the cost-effectiveness studies, our review did not capture the full effect of new treatments in IPF. As the pipeline of available treatments expands, new research will be added to the existing data. We recommend a timely update of this review to capture the influx of new studies and any contemporary research. This will be crucial when informing policy decisions in diagnosis, treatment and palliation of patients with IPF.

5 Conclusion

IPF is a chronic, debilitating condition affecting a growing proportion of the population; predominantly male and the elderly. Our review found evidence of an important health burden of the disease in comparison with HRQoL levels of the general population. Furthermore, our review highlighted an excess cost and resource use for healthcare providers. This confirms IPF as a growing threat for public health worldwide with considerable impact on both patients and healthcare providers.