Prevalence of oxygen desaturation and use of oxygen at home in adults at sea level and at moderate altitude

Chronic hypoxaemia produces several adverse health effects, including pulmonary hypertension, cor pulmonale and death 1, 2. Patients with chronic obstructive pulmonary disease (COPD) and hypoxaemia defined as an arterial oxygen tension (P_a,O2) ≤7.33 kPa (55 mmHg, arterial oxygen saturation (S_P,O2) of 88%), or from 7.33–7.87 kPa (55–59 mmHg) with cor pulmonale or polycythaemia have a reduced survival rate that improves with oxygen therapy 3–5. Altitude is one of the major determinants of hypoxaemia, although age, obesity and lung disease also contribute 1, 2, 6, 7. Prevalence of hypoxaemia in Mexico City, defined as a P_a,O2 <7.33 kPa (55 mmHg), was previously estimated to be between 0.9–3.4% using theoretical models of P_a,O2 decrease with ageing and altitude. 8. A widely used criteria for supplementary oxygen prescription is an S_P,O2 ≤88% 9. The purpose of this study was to estimate the prevalence of oxygen desaturation and its predictors in adults residing at sea level and at a moderate altitude (950 and 2,240 m). Hypoxaemic subjects are also potential users of home oxygen services, an expensive treatment 10 that has to be optimised in the present economic climate of cost containment.

METHODS

Ethical approval was obtained from the ethics committee of the institutions involved in the study (National Institute of Respiratory Diseases, Mexico City, Mexico; University of the Republic, Montevideo, Uruguay; Hospital Universitario de Caracas, Universidad Central de Venezuela, Caracas, Venezuela; and Federal University of Pelotas, Pelotas, Brazil) and all subjects signed a written consent form.

The Proyecto Latinoamericano de Investigacion en Obstruccion Pulmonar (PLATINO) project was planned to estimate the prevalence of COPD in five Latin-American cities, using a multistage cluster sampling, to obtain a minimum of 800 noninstitutionalised subjects per city aged ≥40 yrs. The methods of the PLATINO study have been published 11, as well as the prevalence of COPD in the survey 12.

The subjects answered a respiratory symptoms questionnaire, including sections of the American Thoracic Society (ATS) Division of Lung Diseases questionnaire, the European Community Respiratory Health Survey II, the Lung Health Study and Short Form Health Survey with 12 items (SF-12®) Health Survey 11, 12. Duplicate measurements of body weight (scale HS-301; Tanita Corporation, Tokyo, Japan) and height (208 stadimeter; Seca Corporation, Hamburg, Germany) were made, reporting the averages. Standardisation of measurements were made before the study 13. Spirometric testing (Easy-One; ndd Medizintechnik AG, Zürich, Switzerland) was included, following the ATS recommendations 14 and reference values for Mexican-Americans 15.

S_P,O2 was measured using a pulse oximeter (Onyx 9500; Nonin Medical Inc., Plymouth, MN, USA) in Mexico City (2,240 m above sea level), Caracas (950 m) and Montevideo (35 m), but not in Santiago (Chile; 547 m) or Sao Paulo (Brazil; 800 m). The oximeter was attached to the subject's right index finger and all subjects were seated and breathing room air when the measurement was taken. The average of six measurements taken at 10-s intervals 16 was reported, on the condition that a good pulse signal was present (identified by a green flashing light in the device). The accuracy of five of the pulse oximeters used in the study was tested on all five fingers of one hand of 96 patients whose arterial blood gas was analysed by the pulmonary laboratory of the National Institute of Respiratory Diseases in Mexico City (mean±sd arterial oxygen saturation (S_a,O2) of 87.2±11%). The mean error of S_P,O2 compared with S_a,O2 of a simultaneous arterial blood sample was 0.28% (3.09% sd) and the concordance correlation coefficient between the two measurements was 0.94 (95% confidence interval (CI) 0.93–0.95). The concordance coefficient among instruments was >0.95 and that of each instrument compared with the mean of the five was >0.98.

Altitude was estimated four times in each zone studied in Mexico City and Caracas using an electronic altimeter 42, Altimatic (BARIGO Barometerfabrik GmbH, Villingen-Schwenningen, Germany), and the average reported. In Montevideo, changes in altitude are minimal in the metropolitan area, as it ranges from 1–100 m above sea level.

Sampling design was taken into account in descriptive statistics and regression models by using the “survey” commands of the statistical package. To identify factors associated with S_P,O2, regression models, including age, sex, body mass index (BMI), altitude, current or past smoking, forced expiratory volume in one second (FEV₁), as a percentage of predicted values and the interaction terms of altitude, were fitted with age, BMI, FEV₁ % pred and sex. Oxygen desaturation was defined as an S_P,O2 ≤88%, a common criterion for domiciliary oxygen prescription 9. Impact of oxygen desaturation on health-related quality of life was evaluated by the SF-12® physical score and mental score 17, adjusted by age, sex, lung function, BMI and level of education.

RESULTS

Final participation rate, including questionnaires, pulse oximetry and spirometry, was 73.2, 70 and 78% in Mexico City, Caracas and Montevideo, respectively. Anthropometric data, pulmonary function, oxygen saturation and smoking status of the subjects included in each city are shown in table 1⇓. The final number of participants were: in Mexico City 1,063, of whom 1,052 had S_P,O2 measured (626 females and 426 males); 943 in Montevideo, of whom 843 had S_P,O2 measured (505 females and 338 males); and 1,357 from Caracas all of which had S_P,O2 measured (872 females and 485 males). The mean±sd age was 55.9±12, 60.2±13 and 55.4±11 yrs in Mexico City, Montevideo and Caracas, respectively.

Oxygen saturation in Mexico City (2,240 m above sea level) was lower than in Caracas (950 m) and Montevideo (35 m), but in the three cities it decreased with age (fig. 1⇓). The decrease in S_P,O2 with age increased with altitude, but tended to reach a plateau in older subjects (fig. 1⇓). S_P,O2 also decreased similarly with increasing obesity in the three cities (fig. 2⇓). As the altitude of the city increased, the distribution of S_P,O2 was displaced towards lower saturations, but, in addition, became wider (the sd of the distribution increased significantly) and left skewed, showing a longer tail of lower S_P,O2 values (fig. 3⇓). None of the subjects from Montevideo had an S_P,O2 ≤88%. However, 13 subjects in Caracas (0.96%, 95% CI: 0.6–1.3) and 63 in Mexico City, (6.0%, 95% CI: 4.5–7.5) had an S_P,O2 ≤88%. Prevalence of oxygen desaturation increased with age (table 2⇓). In Mexico City, 10 (0.9%) subjects reported the use of oxygen, five of whom had an S_P,O2 >88%. In Caracas, four (0.3%) subjects reported the use of oxygen, three of whom had an S_P,O2 >88%. Only 8% of subjects with S_P,O2 ≤88% reported the use of oxygen at home in Mexico City and Caracas. Note that the population estimates for the study are stratified by age.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1—

Oxygen saturation by city and age. Population means are depicted as Montevideo (•), Caracas (▴) and Mexico City (▪) with 1.96 se (error bars). Lines and shadows represent the means and 95% confidence intervals, respectively estimated by a quadratic equation that fits the data well. S_P,O2: arterial oxygen saturation.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2—

Oxygen saturation by city and body mass index (BMI). Population means are depicted as Montevideo (•), Caracas (▴) and Mexico City (▪) with 1.96 se (range bars). Lines and shadows represent the means and 95% confidence intervals, respectively estimated by a quadratic equation that fits the data well. S_P,O2: arterial oxygen saturation.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3—

Population distribution of arterial oxygen saturation (S_P,O2) measured by pulse oximetry. The population distribution is shown as the probability function, (the probability that a given S_P,O2 is found in the population). ··········: Mexico City; -------: Caracas; ———: Montevideo.

Variables associated significantly with S_P,O2 in multiple regression models were age, BMI, FEV₁ as a % pred values and altitude. Cumulative smoking expressed in pack-yrs had a small additional negative impact on S_P,O2, but not current smoking. Table 3⇓ shows the multiple regression equations for each city and for the three cities as a group. The latter permits a linear interpolation of S_P,O2 for other altitudes. For example, to obtain a predicting equation for 2,240 m, altitude was replaced by 2.24 (km) in the three cities equation. The resulting equation was very similar, but not identical, to that obtained with Mexico City data, as several altitudes were recorded in the city, and the overall equation was weighted by data from the other two cities. Coefficients for FEV₁ % pred and age were higher in Mexico City than at sea level.

To better describe the adverse impact of ageing and obesity on oxygenation, additional regression models were fitted, including age and BMI as categorical variables, adjusted by FEV₁ % pred and sex. Compared with subjects with BMI ≤25, obesity (BMI>30 kg·m⁻²) was associated with a change in S_P,O2 of −1.3% (95% CI: −0.6– −2.0) in Mexico City, of −0.6% (−0.2– −1.1) in Caracas and of −0.6% (−0.3– −0.8) in Montevideo. Compared with subjects <45 yrs of age, subjects >65 yrs had an S_P,O2 of −2.8% (−2.0– −3.5), −1.7% (−1.5– −2.0) and −0.5% (−0.2– −0.9) in Mexico City, Caracas and Montevideo, respectively.

Significant predictors of hypoxaemia (S_P,O2 ≤88%) in a multivariate logistic regression model were age, obesity, altitude, pack-yrs of smoking and a low FEV₁ % pred. The same predictors applied for S_P,O2 as a continuous variable. Subjects aged >75 yrs had an increased risk of oxygen desaturation (odds ratio (OR): 3.3, 95% CI: 1.6–6.9), as well as subjects with a BMI >30 kg·m⁻² (OR: 1.8, 95% CI: 1.1–3.1), those with medical diagnosis of COPD (OR: 6.1, 95% CI: 1.7–22.0) and residents of higher altitude (OR: 7.5, 95% CI: 4.6 to 12.2 for each km of altitude above sea level). However, only 25% of the variation in hypoxaemia and 38% of variation of S_P,O2 was explained by the multivariate models, emphasising the need to measure S_P,O2.

S_P,O2 correlated with the physical scale but not with the mental scale of SF-12® questionnaire. Subjects with S_P,O2 ≤88% had a lower SF-12® score in the physical scale, even when adjusting by age, BMI, sex, FEV₁ % pred, level of education and altitude above sea level (−2.4 points, 95% CI: −0.6– −4.3).

DISCUSSION

The current study is the first description of the distribution of oxygen saturation in representative samples of large populations. As altitude increases, there is a reduction in mean S_P,O2 but, in addition, the distribution becomes wider and deviates from normality due to an enlarging tail towards lower saturations. Hypoxaemia was estimated previously for the Mexican population, based on the alveolar gas equation, altitude and age 8, although the model did not include obesity or lung disease. However, the model did not predict the low oxygen saturations observed in the studied population. In addition, concordance of measured and predicted oxygen desaturation from the model was very poor for individuals.

An S_P,O2 of 88% corresponds to a P_a,O2 of 7.33 kPa (55 mmHg), a limit frequently used to prescribe home oxygen 3–5 and also separating a group with decreased survival in COPD 3, 4, 5. Approximately 6% of adults >40 yrs in Mexico City (a quarter of million people) and 0.96% in Caracas (∼8,000 subjects) were hypoxaemic by this definition, and would require oxygen at home if the measurement reflects the usual status of the patient. Exposure to altitude is important in Latin-America, either as a permanent residence or in transient visits. For example, in Mexico, half of the population resides >1,500 m 8, and in Peru one-third lives >2,500 m 18.

The adverse effects of age, obesity and lung disease on oxygenation are well known 1, 6, 7, 9. However, altitude was the most important determinant of low oxygen saturation because it was only present in subjects from Mexico City and Caracas, despite the fact that the population in Montevideo was significantly older and had a higher prevalence of airflow obstruction.

In the present study, oxygen was used inappropriately, very little when needed and often without a clear need. Only 8% of hypoxaemic subjects (S_P,O2 ≤88%) reported using oxygen at home, whereas half of the people receiving oxygen at home in Mexico City and three-quarters in Caracas were not hypoxaemic by definition. According to current indications, patients with nocturnal desaturation or hypoxaemia induced by exercise should also receive oxygen, even if the resting S_P,O2 is >88%. However, domiciliary oxygen therapy in subjects with P_a,O2 60–70% or with nocturnal desaturation did not increase survival 19, 20 and, therefore, is a secondary priority for home oxygen services.

Pulse oximeter measurements have good correlation with the gold standard for oximetry in arterial blood samples 21. Although pulse oximeters cannot separate oxyhaemoglobin from carboxyhaemoglobin, S_P,O2 measurements made with pulse oximeters correlate well with saturation of haemoglobin available for gas exchange, even in Mexico City 21, where levels of carboxyhaemoglobin of 1.6% were reported in nonsmokers 22, but can increase several times in smokers. In the present study, although a measurement of carboxyhaemoglobin was lacking, the addition of an indicator of current smoking did not add explaining power to the regression models. The prediction equations based on population samples could be useful to estimate distribution of S_a,O2 and hypoxaemia (S_P,O2 ≤88%) as a function of altitude, age, BMI, and FEV₁. However, the regression models explained only a third of S_P,O2 variability and approximately a quarter of oxygen desaturation, which reinforces the need to measure S_P,O2 and demonstrates the limited understanding of hypoxaemia in populations. Part of this unexplained variation could be reduced with arterial blood gases, but these will be difficult to obtain in population-based samples. Also, better estimates can be expected if population data from cities with altitudes other than those studied are included.

It is important to emphasise that this is a cross-sectional study. Subjects >75 yrs of age had a similar S_P,O2 to subjects aged 65–70 yrs. In previous studies, P_a,O2 also did not decrease with age in subjects >70 yrs 23–25. This is likely to represent a cohort effect, with survivors, and it does not mean that a person of 65 yrs of age should expect a constant S_P,O2 in subsequent years.

The present authors were able to obtain a minimum questionnaire in rejecters, showing that males and the elderly tended to refuse more often than females and younger people, and that the rate of smoking was similar in participants and nonparticipants. This suggests that the reason for refusals was not related to lung disease.

It is necessary to know the impact of hypoxaemia in Mexico City and other cities at similar altitude, in order to properly define the criteria for prescribing oxygen. Currently, the present authors use the criteria that demonstrated improved survival at sea level in subjects with advanced COPD 3–5, 9. The present authors demonstrated an adverse impact of S_P,O2≤88% on the physical scale of the SF-12® questionnaire even after adjusting by altitude, sex, age, FEV₁ and BMI and level of education. This supports the validity of the current definition of hypoxaemia not only close to the sea level but also at higher altitudes. This perhaps supports the current criteria for home oxygen even at altitude. However, information is needed on the long-term impact of hypoxaemia at moderate altitudes on a variety of health indicators, including mortality, quality of life, neuropsychological functioning, pulmonary artery pressure, cardiac function and level of haemoglobin, at least until other indicators of tissue hypoxia are available. Hypoxia is the highest threat to aerobic organisms and, although residence at altitude is likely to have decreased the impact of the so-called “tropical diseases” for centuries, hypoxaemia due to residing at moderate altitudes could now represent a health risk compared with living at sea level.

In conclusion, oxygen desaturation was observed in 6% of the inhabitants of Mexico City and 1% of inhabitants of Caracas aged >40 yrs, but in none of the inhabitants of Montevideo. Only 8% of hypoxaemic subjects in Mexico City and Caracas were using arterial oxygen saturation ≤88%, as measured by pulse oximetry. It is unfeasible to provide home oxygen to 6% of the adult population of Mexico City and, therefore, criteria to prescribe home oxygen more rationally are urgently required.