Long-term domiciliary oxygen therapy (LTOT) is a widely used intervention, although the supporting evidence, based on classic studies such as the Nocturnal Oxygen Therapy Trial (NOTT)1 and the Medical Research Council (MRC)2 trials conducted in the 1980s in COPD patients, is scant and outdated, and the recommendations of these studies have been extended to other diseases that involve hypoxemia. The criteria for indicating this therapy have not been modified by subsequent studies and reviews, nor has new evidence been established,3 and some authors have suggested that the data supporting some indications are limited.

This generalized application of recommendations has also been accompanied by clinical practices in which the initial freedom of prescription has led to some confusion, most often in the form of clinical errors such as incorrect prescriptions and inadequate follow-up, and therapeutic incompliance, understood in a broad sense,4 resulting in misuse of resources and unjustified healthcare expenditure. We have been aware of this problem for many years5–8; however it has recently acquired greater relevance following calls to curb healthcare spending. In other words, it has become more of a management problem than a healthcare issue, although some authors have now begun to question the magnitude of the problem.9

We are a working group with expertise in conducting both local and multicenter home-based LTOT monitoring studies. In view of the growing importance of home respiratory therapies in general and LTOT in particular, we hypothesized that it would be of interest to conduct a new study investigating these factors in order to help define the current state of the problem.

LTOT requires a great deal of patient collaboration due to the number of hours of treatment needed every day if the expected benefits are to be obtained. The degree of adherence to treatment in these patients and the variables that might influence their compliance must be determined. We believe that much of therapeutic non-compliance is associated with the personal characteristics of the patient or the treatment prescribed. The aim of our study was to assess therapeutic adherence among patients receiving LTOT in a health area and the possible influence of active smoking on compliance.

The high level of acceptance of the study on the part of the patients is of note, since only 5 (0.5%) refused to participate. Of the sample studied, 49.1% were men and 50.4% women. Mean age was 82±10.2 years, with a range of between 35 and 104. In total, 46.5% lived in rural areas and 53.5% in an urban environment; 90.9% of the patients lived with others and 8.5% lived alone. Mean time since starting oxygen therapy was 3.3 years.

Analysis of Smoking Habits

Smoking in patients receiving LTOT is contraindicated and may be considered a factor of non-compliance per se. Indirect methods (questionnaire and the presence of signs of smoking in the patient's home) and direct methods (determination of CO in exhaled air) were used to assess the number of smokers.

A total of 459 patients (48.4%) stated in response to the smoking questionnaire that they were smokers or former smokers, and of these, 27 (5.8%) stated they continued to smoke, with an average consumption of 11±8.6cigarettes/day. Four admitted smoking even while receiving oxygen. Based on the first-hand observations made by supply company staff, signs of smoking were detected in the home of 37 patients (8%) of this same group, but not in the remaining 92%. Indications of smoking were found in the homes of all the declared current smokers (ashtrays and cigarette butts); while it obviously cannot be ascertained whether they belonged to the patient or to family members, it seems unlikely that the latter would smoke in front of a patient receiving LTOT. These differences were statistically significant (P<.001). The probability of finding no evidence and the patient being a non-smoker is 122 times greater.

We used the data obtained by indirect methods to study the possible correlations between smoking and patient-dependent factors. The average age of the group of smokers was 67.1±1.5 years versus 83.1±9.6 in the non-smokers group (P<.001). A statistically significant correlation (P=.009) with sex was also found: 74% of smokers were men and 25% women, and the probability of being (or having been) a smoker is 2.9 times greater among men. With regard to place of residence, 70.4% of smokers live in urban areas and 29.6% live in rural areas; while the proportion of smokers is greater in the urban environment, the difference is not statistically significant (P=.077).

Co-oximetry was performed in all study patients, smokers and non-smokers alike, as a direct method of determining tobacco use. Mean values are shown in Table 2. The concentration of CO in exhaled air ranged between 0 and 40ppm. This wide variation may be explained by the fact that the overall study population includes non-smokers, former smokers, and current smokers. Obviously, we found a direct correlation between self-declared current smokers and those in whom signs of consumption and CO in exhaled air were detected (P<.001).

Table 2.

Average Concentrations of CO in Exhaled Air.

	n	Mean	Deviation	P
Overall sample	949	3.6	4.2
Never smokers	490	3.3	0.2	.057
Former smokers	459	3.9	0.2
Self-declared smokers	27	13.9	9.2	<.001
Detected smokers	37	11.4	1.5	<.001

An analysis was made of patients, whether self-declared smokers or not, who had CO levels ≥5ppm, who could be considered likely smokers, and patients with CO ≥10ppm, who are considered smokers. The frequencies are shown in Table 3 and in the graph in Fig. 2. None of the patients who claimed to be never smokers had levels of CO ≥5ppm.

Table 3.

CO Levels in Overall Sample.

Smoker	Co-oximetry		Total	Co-oximetry		Total
	<5ppm	≥5ppm		<10ppm	≥10ppm
No	755	167	922	865	57	922
Yes	5	22	27	10	17	27
Total	760	189	949	875	74	949

Fig. 2.

Number of patients with different levels of CO in the total sample.

(0.06MB).

We also studied possible correlations between CO and patient-dependent factors, but only found a correlation between CO data and the age of the patients (P=.009). Testing the possible correlations between CO values and factors dependent on the characteristics of the oxygen treatment revealed significant correlations only between CO and the hours of use of oxygen therapy. Patients with CO <5ppm received treatment for an average of 12.8h/day; those with CO ≥5ppm received 11h a day (P<.001); and those with CO ≥10ppm received 9.3h a day (P<.001), as can be seen in Fig. 3.

Fig. 3.

Number of hours of compliance with the LTOT according to CO levels in exhaled air.

(0.05MB).

Lack of therapeutic compliance is a widespread phenomenon which affects all types of patients, diseases and treatments, and is the main cause of treatment failure. The multiple factors that determine whether a patient complies or not with treatment are classified as patient-, disease-, and treatment-related factors, and socioeconomic and health system factors.10 Most of these converge in chronic respiratory diseases, which account for the largest group of patients receiving LTOT and those with the greatest degree of non-compliance. COPD,11 after depression, is the disease with the lowest levels of compliance.10

Concern about the problem of compliance in COPD is nothing new. The literature reports highly diverse rates of non-compliance, ranging from 4.2% in the study published in 1978 by Jones et al.12 to 80% in the Antadir study.13 Non-compliance figures have also been published in Spain, ranging from 10% in a study reported by Escarrabil et al.14 in 1985 to 51% reported by Solé et al.5 in 1992. In 2000, Cienfuegos et al.6 reported a rate of non-compliance of 60.5%.

These differences are possibly due to the fact that each study used very different methods to measure compliance, which undoubtedly influences the results. In our study, the rate of non-compliance was 26% when only the survey method was used, but when a direct method was used, the rate was 62.13%, similar to the figures published by Cienfuegos et al.6 This confirms that it is essential to use direct methods to obtain reliable information, since patients are not always truthful about their habits. It is also important that these studies are conducted in patients’ homes, based on surprise visits to avoid false responses. Patients trust the employees of the supplier companies who visit their homes on a regular basis, making them exceptional allies in the conduct of these studies. One study conducted in homes in a health area of Spain in 19918 found a 32.2% rate of compliance, while another multicenter study15 found much higher percentages (79%). Once again, the differences are attributable to the methodology used. In our study, applying the treatment for less than 15h a day was considered non-compliance, as recommended by the different guidelines.

Active smoking is a contraindication for the prescription of LTOT, and can be considered a factor of therapeutic non-compliance, according to the criteria established by the responsible health administrations. In real clinical practice, however, clinicians usually only advise patients against smoking during LTOT, or warn that smoking could lead to suspension of LTOT, a measure that is rarely imposed. However, the proportion of active smokers among these patients is high. In 2000, Cienfuegos et al.6 reported a prevalence in Spain of 11%. In another article published in 2010, Jiménez-Ruiz et al.7 reported a rate of 5.7%.

Given the supposition that smokers receiving oxygen therapy would, in principle, smoke very little, and in order to increase the sensitivity, we decided to establish 2 cut-off points: one at ≥10ppm, which corresponds to certain smokers, and another at ≥5ppm, corresponding to likely smokers. An overall analysis of co-oximetry values shows that 19.9% of the patients have values ≥5ppm, but if the cut-off is set at ≥10ppm, the percentage falls to 7.8% of the total sample.

The truly remarkable finding is that within the sample of current or former smokers, 167 patients who claimed to be former smokers had values ≥5ppm. At this cut-off point, there is a direct correlation between CO values and smoking (P<.001). The likelihood of having values of less than 5ppm is 19.8 times greater in non-smokers. When the cut-off point was set at ≥10ppm in the group who claimed not to smoke, 57 patients had values ≥10ppm, corresponding to certain smokers. At this cut-off point, there is a direct correlation between CO values and smoking (P<.001). The likelihood of having values <10ppm is 25.7 times greater in non-smokers.

A final factor we examined is whether the group of smokers are poorer treatment compliers (in terms of the number of hours a day receiving LTOT). We found that patients with higher values of CO were worse treatment compliers, when both CO >5ppm and CO >10ppm cut-off points were used. This allowed us to conclude that, in addition to their smoking habit, smokers used LTOT fewer hours per day, constituting 2 reasons for treatment failure.

The real benefits of home oxygen, in terms of quality of life and survival according to the degree of respiratory failure,15 remain controversial. However, the fact that many patients receiving LTOT continue to smoke should compel clinicians to step up efforts to help patients quit, and to monitor compliance with LTOT recommendations,16 particularly as patients who continue to smoke receive fewer hours of oxygen per day.

This study has been carried out without external funding, with the collaboration of the company Air Liquide in collecting field data.

The authors declare that they have no conflict of interests with Air Liquide or with the contents of this manuscript.