Article Text
Abstract
Aim Chronic obstructive pulmonary disease (COPD) affects 2–4% of the UK's population and has significant mortality, however prognostication is difficult. Long-term oxygen therapy (LTOT) has been identified by the Gold Standards Framework as a clinical indicator of advancing disease. We hypothesised the ADO index (for ‘Age, Dyspnoea and airflow Obstruction’), which predicts mortality, could be applied to our LTOT population to help identify patients with poor prognosis. We aimed to compare this to the Medical Research Council dyspnoea scale, body mass index, forced expiratory volume in 1 s and gas transfer.
Methods This was a retrospective study on 136 patients with COPD started on LTOT, June 2003 to August 2010. Data were collected from LTOT databases and medical records. Patients’ length of survival was calculated from initiation of LTOT. Patients were grouped by individual parameters. Survival rates at 6 months, 1 year and 4 years were calculated. GraphPad Prism V.5.0 software was used to construct Kaplan–Meier curves and perform log-rank tests.
Results The ADO index discriminated survival at 6 months, 1 year and 4 years, p=0.0027. Low body mass index (<20) was associated with poor prognosis after 1 and 4 years, p=0.0015. Medical Research Council grade predicted survival at 4 years. Diffusion capacity of the lung for carbon monoxide (DLCO)% predicted and forced expiratory volume in 1 s (FEV1)% predicted did not predict survival.
Conclusions This study showed wide variation in survival in a patient population on LTOT. The ADO score could be used as an early trigger for referral to palliative services, thus enhancing end-of-life care, which improves quality of life in COPD. A prospective study of this application would be required to prove this hypothesis.
- Chronic obstructive pulmonary disease
- Prognosis
- Quality of life
Statistics from Altmetric.com
Introduction
Chronic obstructive pulmonary disease (COPD) affects 2% to 4% of the UK's population and is associated with significant morbidity and mortality.1 It is a disease characterised by a steady decline in respiratory function and quality of life, punctuated with acute admissions to hospital with exacerbations.2 Unlike cancer, COPD prognostication is difficult and so patients often receive poor end-of-life care and relatives are often dismayed when an acute exacerbation results in the patient's seemingly untimely death.2 Helping patients understand their illness trajectory is paramount because it empowers them to use supportive care, thus improving their quality of life.3
The Gold Standards Framework is an initiative to improve palliative care within the primary care setting.3 It has identified specific clinical indicators of advanced COPD, which should trigger referral to supportive/palliative care.3 These include the forced expiratory volume in 1 s (FEV1) <30%, Medical Research Council dyspnoea scale (MRC) grade 4/5 and the need for long-term oxygen therapy (LTOT).3 There are several clinical scores used to predict mortality in patients with COPD. The two most commonly used are the BODE index (for ‘Body mass index (BMI), airflow Obstruction, Dyspnoea and Exercise capacity’) and the ADO score (for ‘Age, Dyspnoea and airflow Obstruction’).4 ,5 See online supplementary appendix 1 for an explanation of the variables used within, and scoring of, the MRC dyspnoea scale, BODE index and the ADO score. The ADO score has recently been validated by a large-scale international validation study, which looked at 10 European and American cohorts (n=13 914)
The BODE index includes a 6-min walk test, not always practical to perform, and so we hypothesised that the ADO index could be applied to our LTOT population to better identify patients with poor prognosis. We aimed to compare this prognostic indicator to other parameters including the MRC dyspnoea score, BMI, FEV1% predicted and diffusion capacity of the lung for carbon monoxide (DLCO% predicted).
Methods
Design and setting
A retrospective study was performed on 136 patients with COPD started on LTOT for respiratory failure between June 2003 and August 2010 at the Victoria Infirmary's Respiratory Department, Glasgow. Respiratory failure was defined as resting arterial oxygen pressure (while free of exacerbation) at less than 7.3 kPa, or less than 8.0 kPa with evidence of cor pulmonale. Patients were consecutive; they were started on LTOT following either hospital admission or outpatient clinic assessment.
Data collection
Data were collected within 3 months of LTOT initiation and for this reason some data is missing (eg, results from pulmonary function tests performed before or after this period were discounted). Data were collected from pulmonary function and LTOT databases, as well electronic medical records. Data recorded included date of initiation of LTOT, patients’ age at initiation and total duration of LTOT treatment. Physiological parameters recorded were FEV1% predicted, FEV1 before and after nebuliser use, forced vital capacity, weight, height, BMI and DLCO% predicted along with MRC grade. ADO score at the time of LTOT initiation was calculated. Survival at the end of census or date of death, from electronic case records, was then noted.
Data analysis
Patients’ length of survival was calculated from time of LTOT initiation. We looked at all-cause mortality. Patients were grouped per MRC grade (n=136), ADO score (n=95), FEV1, (n=103), BMI (n=100) and DLCO% predicted (n=45) and we reassessed survival using these parameters. Survival rates at 6 months, 1 year and 4 years were calculated for each parameter. We looked at each individual ADO score, as well as within categories: 2 and 3, 4 and 5 and greater than 6.
Statistical analysis
Statistical analysis was performed using GraphPad Prism 5.0 software to construct Kaplan–Meier Curves and perform log-rank tests, results of which are shown as p values, where p<0.05 is deemed statistically significant. With regard to patient demographics, one-way analysis of variance (ANOVA) was used to determine differences between ADO score groups. SPSS Statistics software (SPSS, Chicago, Illinois, USA) was used to perform Cox regression analysis to adjust for potential confounding variables.
Results
Follow-up and survival data
Minimum follow-up was 1 year unless patients had died. Median follow-up 484 days, mean follow-up 623 days. A total of 35% of the patients were alive at time of census. Minimum survival was 9 days, 25th percentile was 332 days, median was 549 days, 75th percentile was 875 days and maximum survival was 2282 days.
Patient demographics
Please note that the total number of patients studied was 136. However, because data was collected within 3 months of LTOT initiation, some parameters were missing for patients. Therefore, only 95 patients (70%) had sufficient data recorded to enable us to calculate their ADO score. The demographics of these patients are detailed in table 1.
Of the 136 patients, 6 survived beyond 4 years (see figure 1). Three were women, three men. Their mean variables were as follows: age 71 years (SD±8.66), ADO score 4.83 (SD±1.57), MRC grade 3 (SD±0.58), FEV1% predicted 0.56 (SD±0.16), BMI 34 (SD±6.29), DLCO 0.57 (SD±0.06). The mean values for patients surviving less than 4 years were: age 74 years (±9.19), ADO 4.79 (±1.88), FEV1 0.35 (±0.29), BMI 25.7 (±6.97), DLCO 0.39 (±0.17) and MRC grade 3.92 (±1.14) (see figure 2).
Cox proportional hazards model analysis
We quantified the effect of grouping ADO scores with respect to survival using Cox proportional hazards model. We found the HR for patients with ADO 4 and 5 compared to ADO 2 and 3 was 1.44 (CI 0.67 to 3.07), p=0.351. The HR for patients with ADO≥6 compared to ADO 2 and 3 was 2.70 (CI 1.38 to 5.28), p=0.004. We also adjusted for other potential confounders as shown in table 2. BMI was the only variable which was significantly associated with outcome when assessed alongside ADO. We then analysed ADO score alongside BMI, FEV1 and MRC, see table 3. This revealed that ADO remains independently associated with shorter survival after adjusting for other factors known to be associated with a poorer outcome.
ADO index discriminates survival
In the 95 patients with ADO score available there were 55 deaths (58%). ADO index discriminated survival at 6 months, 1 year and 4 years, p=0.0027, see figure 2B,C and table 4. While ADO discriminated survival in this population when individual ADO scores were analysed, due to the relatively small sample size, scores were grouped (figure 2C). Only a single patient had an ADO score of 9, and she survived 125 days.
MRC grade and BMI predict survival at 4 years
In the 100 patients with BMI score available there were 55 deaths (55%). Figure 2D indicates that BMI did not predict survival at 6 months. However, a low BMI (<20) was associated with worse prognosis after 1 year and 4 years, p=0.015. In the 136 patients with an MRC grade available there were 85 deaths (62.5%). MRC grade did not predict survival at 6 months or 1 year, but did predict survival at 4 years, p=0.0118, see figure 2F. There were no patients on LTOT with MRC grade 1.
FEV1% predicted and DLCO% predicted were parameters that did not predict survival in this population.
See figure 2E. Of the 103 patients with FEV1% predicted available there were 56 deaths (54%). In the 45 patients with DLCO% predicted there were 25 deaths (56%). DLCO% predicted does not predict survival, p=0.4132.
Discussion
A wide variation exists in survival of our COPD population on LTOT.
LTOT has been identified as a marker of poor prognosis.3 This study revealed a wide variation of survival in a population on LTOT (see figure 1) and so additional clinical indicators would be of benefit to give a more accurate prediction of mortality in patients being initiated on long-term oxygen therapy.
ADO score analysis
The key finding of this retrospective study is that ADO score appears to discriminate survival probability of patients with COPD requiring LTOT. Patients with a high ADO score, ie, ≥6, have only a 20% chance of being alive at 4 years from initiation of LTOT. This is comparable to the survival statistics for patients with stage 3A non-small-cell lung cancer (NSCLC); 19% to 24% will survive to 5 years.7 Conversely, those with an ADO score of 2–3 have a 65% chance of survival to 4 years. Interestingly, individual parameters within pulmonary function testing did not predict survival, despite FEV1 being a component of the BODE and ADO indices. This is in agreement with previous research looking at predictors of poor prognosis, which state that FEV1 is not the single most important parameter.8 The ADO score was able to discriminate survival at 6 and 12 months after initiation of treatment. In contrast the MRC grade was only able to predict survival at 4 years.
A high ADO score could be used to trigger referral to palliative care.
This study has suggested that, within the subset of patients with COPD receiving LTOT, the ADO index is a valuable tool to predict mortality. This is in agreement with a recent large international study that validated the ADO score as a prognostic tool in the general COPD population, primary care and secondary care.6 It is easy to calculate in general practice and respiratory clinics, unlike the BODE index which requires a 6 min walk test. Along with the other parameters outlined in the Gold Standards Framework,3 the ADO score could act as a trigger to general practitioners (GPs) and respiratory doctors to ask if patients wish to be referred to palliative care colleagues. The score also prompts GPs to review patients’ medications, for example, if patients are known to have a poor prognosis, primary prevention medications such as antiplatelets and lipid-lowering medications are less important.9 Instead, focus can shift to the palliation of symptoms such as dyspnoea, cough, anxiety and fatigue.
Previous studies have outlined inequality between end-of-life treatments received by patients with NSCLC versus patients with COPD. Patients with NSCLC had easily accessible specialist palliative care services whereas those with COPD did not have a comparable system of support.10 Other researchers have looked at specific end-of-life care needs of patients with COPD: diagnosis and disease process, treatment, prognosis, what dying might be like and advance care planning.11 Our study has suggested that we can use the ADO score to give more accurate survival probability in patients on LTOT, compared to using the clinical indicators of advanced COPD as determined by the Gold Standards Framework in isolation. Advance Health Care Directives could then be initiated in appropriate patients. For example, they could prevent the all too common situation where patients with COPD are admitted with severe exacerbations and on-call teams make decisions regarding ceiling of treatment when little information is available about the patients’ previous quality of life or wishes about resuscitation.11
Our study adds to previous studies looking at prognostic factors in patients on LTOT
Another finding from this study was that BMI predicts survival at 1 year and 4 years; this is in agreement with other studies.12 Those patients with BMI >40 had a better outcome. We believe this may be because they have a mixed picture of obesity hypoventilation and COPD, which alters their prognosis. With regard to patients with low BMI, there is currently no formal BMI cut-off point at which patients are referred for dietetic input. Studies have shown that dietary counselling and food fortification resulted in increased BMI and better outcomes for nutritionally at-risk outpatients with COPD.13 Chambellan et al studied the prognostic value of the haematocrit in patients with severe COPD receiving LTOT.15 They found that haematocrit is negatively associated with mortality and morbidity.15
Strengths and limitations of the study
A major strength of the study was the 4-year follow-up period, allowing relatively long-term outcomes to be described. One could argue that using the ADO score was a limitation of the study. Critics of the score argue that it is adversely affected by age; older people with mild COPD in terms of MRC grade and FEV1 are scored worse than younger patients who have severe disease.14 However, the BODE index includes a 6-min walk test, which requires trained personnel and a 30 m hallway, not always practical to perform, especially in the primary care setting.15 The retrospective design of the study was a limitation. Another restraint was the relatively small sample size of only 136 patients from a single centre, giving rise to relatively large 95% CIs when quoting survival probability at 6 months, 1 year and 4 years (see table 4). Furthermore, some patients did not have all parameters recorded for example, gas transfer n=45.
Suggested areas for further research
This study suggests ADO score may be applied in this population to predict all-cause mortality. To extend research in this area, it would be worthwhile performing a comparison between ADO score and the BODE index with regard to survival probability in patients on LTOT. Furthermore, it would be of value to increase the sample size by performing a prospective, multicentred study. It would also be interesting to perform a study to determine whether LTOT reduces hospital admission rates and average length of stay, by collecting and analysing data before and after LTOT treatment. Concordance with LTOT treatment should also be considered in future studies. Further work could examine the interplay between haematocrit and ADO score in predicting mortality. Finally, it would be of interest to perform a prospective study on whether dietetic input improves prognosis and quality of life in patients with COPD on LTOT.
Conclusions
In summary, this retrospective study showed a wide variation in survival in a population on LTOT, identified as a marker of poor prognosis. Our data suggests that ADO score could be used as an early trigger for referral to palliative services. Locally, this would promote referral to the Severe Lung Disease Clinic at our hospital, at which advance care planning and palliative care input can be delivered, thus improving patients’ quality of life.
References
Supplementary materials
Supplementary Data
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 appendix
Footnotes
-
Correction notice This article has been corrected since it was published Online First. A typo was corrected in the Ethics approval statement from ‘itch’ to ‘it’.
-
Contributors DA was responsible for designing and overseeing the entire project, with input regarding design from DR. JM and SB performed the data collection. SL was responsible for the final data analysis, liaising with a statistician at University of Glasgow, and then writing the abstract and paper. DA reviewed the paper prior to submission and is responsible for the overall content as guarantor.
-
Competing interests None.
-
Ethics approval The local institutional review board did not review the study because it was a retrospective analysis on our current practice.
-
Provenance and peer review Not commissioned; externally peer reviewed.
-
Data sharing statement There is a Microsoft Excel database with all of the data stored on it, including unpublished lung function parameters, number of hospital admissions pre-LTOT and post-LTOT. This is available from all of the authors.