This is a cross-sectional analysis of baseline data from the non-randomized controlled MOBIL study (Morbid Obesity treatment, Bariatric surgery versus Intense Lifestyle intervention; Clinical Trials.gov number NCT0027314), the primary aim of which was to compare the effects of lifestyle intervention and bariatric surgery on various comorbidities. The study design, methods, and population have been described in detail [¹¹]. Briefly, 224 adult morbidly obese subjects (BMI ≥40 or ≥35 kg/m² with a weight-related comorbidity), predominantly of European descent, were consecutively screened at our regional tertiary care center. After exclusion of 47 subjects who declined to participate or who had heavy comorbidities, 177 subjects were found potentially eligible for either lifestyle intervention or bariatric surgery. As the aim of this substudy was to explore the specific effects of obesity on blood gases and lung function, 22 subjects with chronic obstructive pulmonary disease (FEV1/FVC <70%) [¹²] and six subjects with restrictive lung function impairment (FVC <80% of predicted value and TLC <80% of predicted value) [¹³] were excluded from further analyses. The remaining 149 subjects were defined as having normal lung function and were included in the analysis.

All participants gave informed written consent before enrolment. The study was approved by the Regional Ethics Committee for Medical Research (S-05715).

All subjects were examined by a physician. Demographic data and medical history, including smoking habits (recorded as pack-years), and use of bronchodilators were recorded. Height, weight, neck circumference (NC), waist circumference (WC), and hip circumference (HC) were measured with subjects in an upright position wearing light clothes and no shoes. Indexes were calculated for BMI (weight divided by the square of height, kilograms per square meter (kg/m²)) and WHR (WC divided by HC). NC was measured at the cricoid cartilage and WC at the level midway between the lowest rib margin and the iliac crest.

Blood gas samples were taken from the radial artery after at least 5 min of rest and with the subjects in a sitting position. The sampling was carried out by either a physician (AMG) or two experienced nurses. For analyses we used an ABL 735 Radiometer (Copenhagen, Denmark) calibrated in accordance with the manufacturer's specifications. Arterial blood gas values for pO2 < 10.7 kPa (80 mmHg) and for pCO2 > 6.0 kPa (45 mmHg) were defined as abnormal [¹⁴].

Lung function measurements included dynamic spirometry, static lung volumes, and gas diffusion capacity. All tests were carried out by two experienced nurses according to the guidelines recommended by the ATS-ERS task force [¹⁵–¹⁷] and with subjects sitting in an upright position wearing a nose clip. The Jaeger Master Lab (Eric Jaeger, Wurzburg, Germany) was used in all tests and was calibrated daily using a 1-L syringe. Recorded variables were: FVC, FEV1, FEV1/FVC, diffusion capacity for carbon monoxide (DLCO), DLCO divided by alveolar lung volume (DLCO/VA), TLC, VC, inspiratory capacity (IC), intrathoracic gas volume (ITGV), ERV, and residual volume (RV).

The reference values for FVC, FEV1, DLCO, TLC, VC, IC, ITGV, and RV are those recommended by the European Respiratory Society (ERS) [¹³]. These reference values have been derived from studies of healthy, non-smoking adults of European descent [¹³]. However, since the ERS offers no reference values for ERV, the Jaeger Master Lab reference values were used instead. Sixteen lung volume tests were rejected (12 due to technical problems and four due to compliance problems), and one single breath test was rejected (due to non-compliance). All tests were carried out between 9 and 10 A.M. No bronchodilator drug was allowed prior to lung function testing.

Independent samples t test or Mann-Whitney test were used for the comparison of continuous data, whereas Chi-square test was used for categorical data. Spearman's correlations were calculated to assess bivariate relationships between anthropometric measures, blood gases, and lung function tests. Multiple linear regression analyses were used to assess the effect of each anthropometric measure on arterial blood gases after adjustment for confounding factors (age, gender, pack-years). We fitted eight separate multiple linear regression models with either pO2 or pCO2 as the dependent variable, and either BMI, NC, WC, or WHR as the independent variable adjusted for confounders. No violation of the assumptions of normality or multicollinearity was revealed in the models. Two-tailed P values are reported, with P values below 0.05 considered statistically significant. Semi-partial (part) correlation coefficients were squared in order to estimate the percentage of total variance in the dependent variable explained by a given independent variable. The analyses were carried out using SPSS 16.0 (SPSS, Chicago, IL).

The main and novel findings of this study are that both measures of overall (BMI) and central obesity (WC, NC) were significantly associated with lower pO2 and higher pCO2 among morbidly obese subjects with normal lung function. In addition, BMI, NC, and WC correlated negatively with ERV.

Our results are partly in contrast to those of Zavorsky et al. [⁹] who found that increasing WHR, but not BMI, was associated with detrimental changes in blood gas values. This discrepancy might have several explanations. Firstly, the previous study was small (n = 25) and included a heterogeneous sample of subjects with either normal (n = 18) or abnormal (n = 7) lung function, but the analysis was not stratified according to lung function. Conversely, we examined 149 subjects, all with normal lung function. Secondly, in contrast to our study, clinically relevant confounding factors were neither addressed nor adjusted for in the previous study. Notably, we too could demonstrate statistically significant associations between WHR and blood gases in the univariate analyses, but these associations did not remain significant after adjustment for gender, age, and pack-years.

Our findings partly confirm the preliminary results from a cross-sectional study of 76 morbidly obese Chinese subjects that showed a significant association between BMI and blood gases [¹⁸]. Our results extend those findings to be valid in white morbidly obese patients independent of confounding factors. No measure of central obesity was addressed in the Chinese study.

Another more recent cross-sectional study of 150 morbidly obese Chinese adults with normal lung function demonstrated that only WC—in contrast to BMI and WHR—was associated with reductions in FEV1, FVC, TLC, and VC [¹⁹]. However, this study did not address the relationship between anthropometric measures and arterial blood gases.

Abdominal fat raises intra-abdominal pressure causing cephalad movement of the diaphragm, and thus reducing chest volume. Morbidly obese subjects have an abnormally high alveolar-to-arterial oxygen partial pressure difference and reduced pO2 [⁹, ¹⁰], which reflects a ventilation–perfusion mismatch as the lower parts of the lungs are underventilated and overperfused in obese subjects. Accordingly, profound weight loss after bariatric surgery has been associated with increased pO2 [⁷, ¹⁰]. Both WC and BMI reflect the amount of abdominal fat [²⁰], and consequently increasing BMI and WC may be associated with atelectases and ventilation-perfusion mismatching, which may have an impact on blood gases.

In accordance with a previous study of morbidly obese patients, most lung function values were within predicted range whilst the mean ERV was less than 50% of predicted value [²¹].

Further, ERV was significantly correlated with BMI, NC, and WC. A decrease in ERV may lead to a decrease in FRC which in turn may fall below closing volume. Thus, morbidly obese subjects breathe near their residual volume. The collapse of small airways leads to increasing V/Q mismatch and hypoxia as described above. The relationship between reduced ERV and hypoxia in morbidly obese subjects has previously been shown [⁷]. In a study aimed to assess the effects of BMI on lung volumes [²²], FRC and ERV were found to decrease exponentially with increasing BMI, and the greatest effects were on subjects with BMI >30 kg/m². Notably, both FRC and ERV were significantly lower in moderately obese women and men (BMI, 30–35 kg/m²) than in their normal weight counterparts (BMI, 20–25 kg/m²) of comparable age. In this way, even moderate obesity lead to decreased ERV which may affect blood gases.

Notably, although we excluded patients with abnormal lung function from the present analysis, one out of four patients (27%) had been prescribed bronchodilator drugs due to asthma as diagnosed by their physician. This finding is in line with a previous study which reported a prevalence of asthma of 24% among morbidly obese subjects [²¹]. Several biological mechanisms may explain the relationship between asthma and obesity including those referred above [²³].

Our study demonstrates that in morbidly obese subjects, increasing levels of obesity are associated with worsening of blood gases. This could imply that clinicians should use spirometry and blood gas analyses to identify morbidly obese patients at particularly high risk for post-operative complications—for instance after bariatric surgery. Consequently, it might be wise to measure lung function and blood gases routinely in this patient group.

Our study has limitations. Firstly, the cross-sectional design is unsuitable for establishing a cause–effect relationship. Secondly, our results may not be valid in non-white populations. Finally, we cannot exclude the possibility that the referral of patients to a tertiary care center might have introduced a sampling bias which may limit the generalizability of our results.

Our study of white morbidly obese subjects with normal lung function, demonstrates that increasing levels of BMI, WC, and NC are significantly associated with decreasing pO2 and increasing pCO2 even after adjustments for age, gender, and pack-years. In addition, both central and overall obesity are associated with low ERV.

Values are given as mean (SD) or n (%)

Except for blood gases, values are referred as “percent predicted” and given as mean (SD)

Each anthropometric measure was adjusted for age, gender, and pack-years in eight separate multiple linear regression analyses

BMI denotes body mass index, NC neck circumference, WC waist circumference, WHR waist-to-hip ratio