The Power of Annual Cost-of-Illness in Predicting COPD Survival at 3 Years

FEV1: Forced Expiratory Volume in 1 second

FEV1/FVC: Forced Expiratory Volume in 1 second/Forced Vital Capacity

RV: Residual Volume

DLco: Lung Diffusion Capacity for CO

BMI: Body Mass Index

Chronic Obstructive Pulmonary Disease (COPD) is a complex and progressive condition which is characterized by a dramatic, ever increasing socio-economic impact [1-4]. Identifying the mortality risk in COPD patients still represents a crucial issue in the strategic management of this condition and the assessment of accurate predictors would support more effective interventions aimed at improving outcomes. 

Actually, the sensitivity of clinical signs and of usual parameters of lung function (mostly, FEV1 and FEV1/FVC) in predicting death still is variable, even if they prove able to better define COPD long-term outcomes when integrated into multidimensional scores [5-13]. 

The role of the annual economic impact was absolutely neglected from this point of view and only in one study the annual cost was included in a composite score [14]. 

Aim of the present study was to assess the power of COPD annual cost-of-illness in predictive survival at three years. The relationships existing between all the parameters collected and survival were also investigated.

The patients’ sample was automatically extracted from the Institutional Data Base. Criteria for selection were:

• • Subjects of both genders aged ≥ 40 years;
• • Diagnosis of COPD;
• • Yearly availability of at least one complete electronic clinical record for four consecutive years, such as: During the twelve months preceding the visit T₀ and also for the three following years;

The variables to collect at T₀ were: Age; gender; BMI; smoking habit; clinical data (namely: BODE Index; MRC Dyspnea score; Charlson Comorbidity Index; n. hospitalizations and their duration; n. exacerbations requiring antibiotic and/or systemic steroids assumption; complete lung function (namely: FEV1 in L and in % predicted; FEV1/FVC %; RV % predicted; DLco % predicted; FEV1 short-term reversibility after salbutamol 400mcg; 6’ walking test (in metres) and the annual cost calculated over the previous twelve months. 

All-cause mortality and the length of survival were assessed by careful review of the death certificate delivered by the registry office of the Institution.

Parametrical, non-parametrical tests and anova were used for comparing means±SD; 95% CI was also calculated. Linear regression was used in order to assess the total annual cost calculated at T₀ and the length of survival over the following three years. The power of each variable considered at T₀ in predicting mortality at three years (such as: Patients who will survive and patient who will decease during the study period) was also calculated and represented by web diagrams including the means value for each variable together to the corresponding statistical significance. The lower accepted limit of significance was p<0.05. 

Selected subjects were included only after their informed consent to the study, which was approved by the Institutional Human Review Board of the Institution. 

The present study was a spontaneous research, without any founding. Authors declare no conflict of interest.

The total sample consisted of 275 COPD patients who satisfied the selection criteria over the period 2008-2012. General characteristics of the whole sample are reported in table 1. 

Table 1: Means±SD calculated for each variable at T₀ in the whole sample (n=275). 

The overall mortality rate was 40.4% over three years (n=111; n=12, 47 and 52 subjects deceased yearly). The three-year survival ranged 1024-755 days in patients deceased, respectively. 

At T₀, the lung function profile of subjects still surviving after three years (n=164) proved significantly different from that of patients who will decease in the same period. The two clusters of patients were different in terms of mean age; FEV1(L); RV% predicted; DLco % predicted; 6’ walking test; and Charlson comorbidity index (all p<0.001), but not in terms of FEV1% predicted and FEV1/FVC% (all p=ns) (Figure 1). In particular, patients who will die showed a more relevant emphysema component and were characterized by the highest annual cost, despite their shorter survival.

Figure 1: The web diagram summarizing the power of each lung function variables assessed at T₀ in discriminating patients who will survive from those who will decrease over the three following years of the survey. Values reported correspond to the mean value calculated for each variable. 

All economic components of cost were highly discriminant, independently of gender (Figure 2). Mean total COPD cost was 3,290.7 [95% CI: 2539.9; 4051.2], while it was € 1,161.0 [95% CI: 968.4; 1,353.6] in survivors and 6,158.9 [95% CI: 5,508.0; 6,809.8] in those who will die over the same period, respectively (p<0.001). The hospitalization cost impacted for 78.2% of the total annual cost in the latter subjects and their absolute cost proved 6-fold higher than that of survivors. Moreover, total annual cost showed the highest regression with survival (r = -0.63).

Figure 2: The web diagram summarizing the power of each determinant of the annual cost assessed over the twelve months before T₀ in discriminating patients who will survive from those who will decrease over the three following years of the survey. Values reported correspond to the mean value calculated for each variable.

The mortality risk in COPD patients had been investigated in several studies in the last decades by means of a single parameter [15-19], by complex and composite clinical assessments [5-13] and by the recent multidimensional scores [20-23]. 

The different approaches to the prediction of COPD survival progressively improved the overall accuracy. Nevertheless, a wide variability still persists, mainly due to the variables originally chosen for the studies and frequently affected by the primary aim of these studies (i.e. pharmacological, epidemiological, etc). 

FEV1 represented the simplest and then the first and the best accepted variable used to predict COPD mortality since long ago, even if it cannot reflect exhaustively the huge complexity of changes related to COPD survival [6]. 

Multidimensional scores improved the sensitivity of the assessment of mortality risk in COPD patients as, though with different specificity, these instruments valued several factors affecting COPD clinical severity and prognosis [6,9,12]. 

Differently from symptoms subjectively claimed by patients (directly reported or reported via generic and specific questionnaires) and from usual simple lung function indices, the annual cost-of-illness represents an objective and comprehensive parameter which is able to summarize dynamically all the critical issues and the real-life severity and progression of COPD. The major components of the cost depend on the exacerbation and the hospitalization rates and duration and on the extent of drug utilization: These three components account for > 80% of total annual cost. As a consequence, it is absolutely obvious that higher the annual cost, higher the severity and the progression of the disease should be, because characterized by frequent exacerbations, more frequent hospitalizations and requiring a larger use of drugs. 

Data of the present study are confirming that the annual cost-of-illness should be regarded as the comprehensive parameter most strictly related to the length of survival over a three-year temporal horizon in COPD: Much more related than any other multidimensional score. Only one paper published a few years ago suggested the challenging concept of the convenience of including the economic impact in a multidimensional score with the aim to assess the severity of COPD [14]. 

Moreover, data of the present study are also emphasizing the crucial role of proper lung function indices in order to assess the mortality risk. In fact, when the investigation of different aspects of lung function is based on more suitable and specific lung function tests (i.e. RV, DLco), the relationship with the length of survival immediately becomes clear, because emphysema components can be basally assessed. Not by chance, these patients are the same that induce the highest annual cost despite their shorter survival. 

Limitations and strengths of the study 

The sample consisted of COPD patients referring to a single Lung Unit, even if the majority of patients were belonging to different Italian regions. 

The careful characterization of patients in clinical and physiological terms and the careful assessment of primary outcomes (annual cost and mortality) provide strength to the study, prospective in nature. 

Despite the costs here considered may be regarded as corresponding to those of the Veneto Region, the costs were uniquely derived from the national DRG tariffs and prices [24] and the basic methodology adopted for the present survey might be adopted in all Countries, independently of the model of their health system. On the other hand, these costs result absolutely comparable to those of other developed Countries and have the advantage of having been based on real-life recent costs of health care. 

When compared to the mean duration of COPD, the length of the survey (3 years) might appear relatively short, even if several big studies have the same duration. Anyhow, health care providers and public decision makers ask for and need data at medium term in order to modulate dynamically their interventional plans of chronicity containment and management.

The present study confirms that, even if easy to obtain, the most used indices of lung function (namely, FEV1% predicted and FEV1/FVC%) show the lowest power in predicting survival at 3 years, because characterized by a low sensitivity and specificity. 

RV% predicted; DLco% predicted proved much more strictly related because more specific in defining the overall lung function pattern and much more appropriate for discriminating emphysema components in COPD (such as, a particular phenotype). 

Finally, the total annual cost-of-illness confirms the most sensitive global predictor of mortality in COPD. 

In other words, data are supporting the high predicting value of the “functional and economic phenotyping” in terms of COPD patients’ survival.