Diseases of the pleural space (pneumothorax, pleural effusion, tumors, and diaphragmatic hernia) are relatively common in small animal clinical practice.1 They can quickly become life-threatening.2 In a retrospective study2 of pleural effusions of various causes, almost one-third of the sick dogs died during investigation or after initial evaluation. This highlights the importance of early detection to correctly adjust diagnostic investigations and, if necessary, carry out appropriate emergency measures.
Some clinical signs, such as muffled heart sounds, are evocative of pleural diseases. These are, however, inconsistently recognized.2 Thoracic radiographs are an effective tool for the diagnosis of pleural diseases. However, patients are sometimes too unstable to immediately undergo such an imaging procedure. It would therefore be useful to identify a reliable and easily recognizable criterion that could be used to rapidly detect pleural space diseases.
In dogs, respiratory patterns associated with pleural diseases have been reproduced via experimental increases in pleural pressure.3,4 Increased pleural pressure is known to decrease inhibition of inspiratory intercostal muscle activity. Results of previous studies3,4 indicate that when these muscles work harder, a greater elevation of the ribs during inspiration occurs. Increased inspiratory effort seems to be strong enough to aspirate the abdominal viscera toward the chest, resulting in sunken flanks when the ribs elevate and bulging flanks when the ribs depress. Various names have been assigned to this particular respiratory pattern,5–12 the most common being PB or discordance. As the inspiratory efforts increase, the respiratory movements are marked. Therefore, PB is usually very easy to recognize. However, to the best of our knowledge, the clinical relevance of PB associated with pleural diseases has never been assessed in small animals. The objective of the study reported here was therefore to retrospectively assess and compare the occurrence of PB and pleural diseases in a large population of dogs and cats with dyspnea.
Materials and Methods
Study design and population—Case records of dogs and cats that had undergone a complete physical examination leading to the diagnosis of dyspnea between January 2001 and October 2009 at the National Veterinary Schools of Toulouse and Alfort were retrospectively identified by use of a computerized medical database search with the coding term dyspnea. Identified records were then individually reviewed by 1 investigator (KLB).
Signalment (age, sex, breed [cats], and morphotype [dogs]) and breathing pattern at the time of diagnosis (ie, presence or absence of PB) were recorded. For dogs, morphotype was defined as the interaction between body weight (< 10 kg [22 lb], 10 to 25 kg [22 to 55 lb], or > 25 kg) and cranial morphology (mesocephalic, brachycephalic, or dolichocephalic). For cats, breed was considered as the major morphological characteristic, and body weight was not taken into account.
Diagnosis of pleural diseases (eg, diaphragmatic hernia, pleural effusion, and pneumothorax) was made on the basis of the results of thoracic radiography showing at least one of the following abnormalities: pleural fissure lines, retraction of the lungs from the thoracic walls by a soft tissue or gas opacity, loss of the diaphragmatic line, or air-filled visceral structures in the thoracic cavity. All of the radiographs were interpreted by a senior clinician at time of diagnosis. Animals were then classified into 2 groups, according to the presence or absence of pleural diseases. Patients with missing data (eg, no thoracic radiographs) or equivocal characterization of PB were not included in the study.
Statistical analysis—Statistical analyses were performed with a commercially available software package.a Occurrences of PB and pleural diseases in dogs and cats were compared with χ2 analysis. Stratification by species was then used for the full analysis.
Association between PB and pleural diseases was examined by use of univariate and multivariate logistic regression analyses. Odds ratios were taken as an approximation of relative risk and a measure of the strength of the association and were reported with their 95% CIs. Age was converted into 4 categories defined by the quartiles of the observed distribution because its relationship with PB was not linear. Age, sex, feline breed, canine weight, canine morphology, and hospital were considered to be potential confounders. Variables were considered to be potential confounders if they could be linked to both PB and pleural diseases. Interactions between these variables were tested. None of the examined factors were significantly associated. For dogs, the interaction between weight and cranial morphology (ie, morphotype) was tested to identify a relationship with PB because this allowed us to define specific subgroups of dogs that shared the same morphological characteristics. Although a significant interaction between weight and cranial morphology was not identified, we elected to include morphotype in the canine model because canine morphotype made sense from a clinical point of view. The Hosmer-Lemeshow test statistic was used to determine the goodness of fit for the model.
Signalment of animals with and without PB were assessed and compared. For dichotomous and categorical variables (sex, feline breed, and canine morphotype), a χ2 analysis or a 2-tailed Fisher exact test was used, as appropriate. For age, the Mann-Whitney test was used because its distribution deviated from the normal distribution (graphic evaluation and Shapiro-Wilk test). For all the analyses performed, tests used were 2 tailed and values of P < 0.05 were considered significant.
Results
Study population—The study population was composed of 389 dyspneic animals (195 dogs and 194 cats) from both centers (255 from Alfort and 134 from Toulouse). Occurrence of PB (53/195 [27.2%] dogs; 125/194 [64.4%] cats) and pleural diseases (39/195 [20.0%] dogs; 91/194 [46.9%] cats) differed significantly (P < 0.001) between dyspneic dogs and cats, justifying subsequent stratified analyses by species.
Signalment of dogs and cats with PB did not differ significantly from signalment of dyspneic dogs and cats without PB (Tables 1 and 2). Patient characteristics according to the presence or absence of pleural diseases were summarized (Tables 3 and 4). Canine weight was the only significantly (P = 0.03) different characteristic between animals with or without pleural diseases.
Signalment of dogs (n = 195) with a recorded diagnosis of dyspnea examined at the National Veterinary Schools of Alfort and Toulouse (France) between January 2001 and October 2009 with and without PB.
| Variable | PB (n = 53) | No PB (n = 142) | P value* |
|---|---|---|---|
| Age (y)† | 8.0 (0.16–16.0) | 9.5 (0.25–16.0) | 0.81 |
| Sex ratio‡ | 0.77 (0.43–1.31) | 1.29 (0.93–1.82) | 0.11 |
| Body weight (kg)§ | 0.45 | ||
| < 10 | 25 (47.2) | 57 (40.1) | |
| 10–25 | 11 (20.7) | 42 (29.6) | |
| > 25 | 17 (32.1) | 43 (30.3) | |
| Morphology§ | 0.40 | ||
| Mesocephalic | 41 (77.4) | 100 (70.4) | |
| Dolichocephalic | 4 (7.5) | 8 (5.6) | |
| Brachycephalic | 8 (15.1) | 34 (24.0) |
Assesses presence or absence of significant differences for each characteristic between the 2 groups of dogs.
Values are median (range).
Male to female (95% CI).
Values are number (%).
Signalment of cats (n = 194) with a recorded diagnosis of dyspnea examined at the National Veterinary Schools of Alfort and Toulouse (France) between January 2001 and October 2009 with and without PB.
| Variable | PB (n = 125) | NoPB (n = 69) | P value* |
|---|---|---|---|
| Age (y)† | 7.1 (0.12–19.0) | 7.1 (0.1–17.0) | 0.93 |
| Sex ratio‡ | 1.16 (0.81–1.65) | 0.97 (0.60–1.57) | 0.56 |
| Breed§ | 0.95 | ||
| Domestic | 108 (86.4) | 62 (89.9) | |
| Siamese | 6 (4.8) | 3 (4.4) | |
| Persian | 4 (3.2) | 2 (2.9) | |
| Maine Coon | 2 (1.6) | 1 (1.4) | |
| Other‖ | 5 (4.0) | 1 (1.4) |
Other breeds include 1 Chartreux, 1 Oriental Shorthair, 1 Abyssinian, 1 Russian Blue, 1 British Shorthair, and 1 Birman.
See Table 1 for remainder of key.
Signalment of the dyspneic dogs (n = 195) in Table 1 with and without pleural disease.
| Variable | Pleural disease (n = 39) | No pleural disease (n = 156) |
|---|---|---|
| Age (y)† | 7.0 (0.25–16.0) | 10.0 (0.16–16.0) |
| Sex ratio‡ | 0.75 (0.37–1.44) | 1.22 (0.90–1.69) |
| Body weight (kg)§ | ||
| < 10 | 12 (30.8) | 70 (44.9) |
| 10–25 | 8 (20.5) | 45 (28.8) |
| > 25 | 19 (48.7) | 41 (26.3) |
| Morphology§ | ||
| Mesocephalic | 31 (79.4) | 110 (70.5) |
| Dolichocephalic | 4 (10.3) | 8 (5.1) |
| Brachycephalic | 4 (10.3) | 38 (24.4) |
See Table 1 for key.
Signalment of the dyspneic cats (n = 194) in Table 2 with and without pleural disease.
| Variable | Pleural disease (n = 91) | No pleural disease (n = 103) |
|---|---|---|
| Age (y)† | 5.0 (0.12–19.0) | 7.0 (0.10–19.0) |
| Sex ratio‡ | 1.13 (0.74–1.73) | 1.04 (0.71–1.54) |
| Breed§ | ||
| Domestic | 80 (87.9) | 90 (87.4) |
| Siamese | 3 (3.3) | 6 (5.8) |
| Persian | 2 (2.2) | 4 (3.9) |
| Maine Coon | 3 (3.3) | 0 (0.0) |
| Other‖ | 3 (3.3) | 3 (2.9) |
See Tables 1 and 2 for key.
Association between PB and pleural diseases in dogs—Occurrence of pleural diseases was significantly higher in dyspneic dogs expressing PB (26/53 [49.1%]), compared with those that did not express PB (13/142 [9.2%]; P < 0.001). Paradoxical breathing as a predictor of pleural diseases in dyspneic dogs had an Se of 0.67, Sp of 0.83, PPV of 0.49, and NPV of 0.91. The crude OR for the comparison between pleural diseases and PB was 9.6 (CI, 4.1 to 22.7), indicating that PB was highly associated with an increased risk of pleural diseases in dogs. The OR adjusted for age, sex, hospital, weight, morphology, and interaction between weight and morphology values (Table 5) was even higher: 12.6 (CI, 4.6 to 31.2). Among all potential confounders, only hospital was an actual confounder because the OR between PB and pleural diseases changed when this variable was removed from the logistic model. Indeed, PB was more likely to be recorded at the National Veterinary School of Alfort than at Toulouse (adjusted OR, 4.0; CI, 1.7 to 9.6). The goodness-of-fit statistic (Hosmer-Lemeshow) suggested that the data adequately fit the model (P = 0.44).
The pleural diseases identified in dogs with PB included pleural effusions (22/26 [85%]), pneumothorax (3/26 [12%]), and pyothorax (1/26 [4%]). When PB was identified without pleural diseases, etiology of dyspnea was as follows: acute pulmonary edema (5/27 [19%]), interstitial pneumopathies (4/27 [15%]), severe abdominal effusion (4/27 [15%]), upper airway obstruction (3/27 [11%]), lower airway obstruction (3/27 [11%]), bronchopneumonia (2/27 [7%]), pulmonary thromboembolism (1/27 [4%]), lung tumor (1/27 [4%]), severe hepatomegaly (1/27 [4%]), gastric volvulus (1/27 [4%]), and severe peritonitis (1/27 [4%]). For 1 (4%) dog, no cause was identified.
Association between PB and pleural diseases in cats—Occurrence of pleural diseases was significantly (P < 0.001) higher in dyspneic cats expressing PB (82/125 [65.6%]), compared with those that did not express PB (9/69 [13.0%]). Paradoxical breathing as a predictor of pleural disease in dyspneic cats had an Se of 0.90, Sp of 0.58, PPV of 0.66, and NPV of 0.87. The crude OR between pleural diseases and PB was 12.7 (CI, 5.5 to 31.6). The OR for pleural diseases after age, sex, hospital, and breed adjustment was higher than in univariate analysis (14.1; CI, 6.0 to 33.5) and remained highly significant (P < 0.001). Among all potential confounders, only the hospital was an actual confounder because the OR between PB and pleural diseases changed when this variable was removed from the logistic model (Table 6). As for dogs, PB was more likely to be recorded at the National Veterinary School of Alfort than in Toulouse (adjusted OR, 3.7; CI, 1.6 to 8.3). The goodness-of-fit statistic (Hosmer-Lemeshow) suggested that the data adequately fit the model (P = 0.41).
Multivariate analysis of relationships between pleural diseases, potential confounders, and PB in the same dyspneic dogs as in Table 1.
| Variable | Adjusted OR | 95% CI | P value |
|---|---|---|---|
| Pleural disease | 12.6 | 4.6–31.2 | < 0.001 |
| Age (y)* | |||
| 5–9 | 2.3 | 0.7–7.2 | 0.15 |
| 10–12 | 1 | 0.3–3.2 | 0.97 |
| 13–16 | 1.9 | 0.6–6.6 | 0.30 |
| Female† | 1.5 | 0.7–3.3 | 0.35 |
| Body weight (kg)‡ | |||
| 10–25 | 0.4 | 0.1–1.2 | 0.10 |
| > 25 | 0.3 | 0.1–1.0 | 0.05 |
| Morphology§ | |||
| Brachycephalic | 0.4 | 0.1–1.4 | 0.15 |
| Dolichocephalic | 1.5 | 0.1–23.6 | 0.76 |
| Morphotypeठ| |||
| Medium brachycephalic | 6.1 | 0.4–94.6 | 0.19 |
| Large brachycephalic | 12.2 | 0.5–299.4 | 0.13 |
| Medium dolichocephalic | 0.5 | 0.01–26.6 | 0.75 |
| Large dolichocephalic | 2.3 | 0.05–108.7 | 0.67 |
| National Veterinary | 4 | 1.7–9.6 | 0.002 |
| School of Alfort‖ | |||
Reference category: 0 to 4.
Reference category: male.
Reference category: < 10 (small).
Reference category: mesocephalic.
Reference category: National Veterinary School of Toulouse.
Multivariate analysis of relationships between pleural diseases, potential confounders, and PB in the same dyspneic cats as in Table 2.
| Variable | Adjusted OR | 95% CI | P value |
|---|---|---|---|
| Pleural disease | 14.1 | 6.0–33.5 | < 0.001 |
| Age (y)* | |||
| 3–5 | 0.9 | 0.3–2.8 | 0.92 |
| 6–12 | 0.7 | 0.3–1.8 | 0.47 |
| 13–19 | 1.1 | 0.4–3.1 | 0.83 |
| Female† | 0.9 | 0.4–1.8 | 0.72 |
| Breed‡ | |||
| Siamese | 1.5 | 0.3–8.2 | 0.62 |
| Persian | 2.1 | 0.3–15.5 | 0.47 |
| Maine Coon | 0.1 | 0.01–1.8 | 0.13 |
| Other§ | 2.8 | 0.2–31.7 | 0.40 |
| National Veterinary | 3.7 | 1.6–8.3 | 0.002 |
| School of Alfort‖ | |||
Reference category: 0 to 2.
Reference category: male.
Reference category: domestic.
Other breeds include 1 Chartreux, 1 Oriental Shorthair, 1 Abyssinian, 1 Russian Blue, 1 British Shorthair, and 1 Birman.
Reference category: National Veterinary School of Toulouse.
Pleural diseases identified in cats with PB were as follows: pleural effusions (54/82 [66%]), pneumothorax (23/82 [28%]), diaphragmatic hernia (4/82 [5%]), and pyothorax (1/82 [1%]). When PB was recognized without pleural diseases, dyspnea origins were as follows: lower airway obstruction (12/43 [28%]), acute pulmonary edema (7/43 [16%]), lung tumors (6/43 [14%]), upper airway obstruction (4/43 [9%]), interstitial pneumopathies (4/43 [9%]), bronchopneumonia (2/43 [5%]), pulmonary thromboembolisms (2/43 [5%]), and ketoacidosis diabetes (1/43 [2%]). For 5 cats (12%), no cause was identified.
Discussion
The results of the present study indicate that PB is strongly associated with pleural diseases in dyspneic dogs and cats (adjusted OR > 10 for both species in multivariate analyses). Age, sex, feline breeds, and canine morphotypes in patients with PB were not significantly different from those of other dyspneic animals. Reports of PB in the veterinary literature are rare. Therefore, little is known about this particular breathing pattern that is frequently classified as a restrictive dyspnea. However, this relationship between PB and pleural diseases has been demonstrated in human medicine. For example, PB is reported to be a key symptom of diaphragmatic paresis in human patients,8–13 although spontaneous diaphragmatic paresis was not observed in the present study. In people, PB has also been shown to be associated with neurologic diseases,12,14 diaphragmatic myopathy,9 flail chest,10 chronic obstructive pulmonary disease,15 upper airway obstruction,16,17 severe abdominal distension,18 and pleural effusion.19 In the present study, PB was mainly associated with pleural effusion, pneumothorax, pyothorax, and diaphragmatic hernia, which are considered to be the most common pleural diseases in both cats and dogs.1 This was particularly true in cats, which expressed PB in almost 90% of patients with pleural diseases.
However, similar to human patients, PB was also seen in our study in patients with nonpleural diseases (27/156 [17%] and 43/103 [42%] of dyspneic dogs and cats with nonpleural diseases, respectively). Causes of dyspnea in these patients included upper airway obstruction, pulmonary diseases (interstitial pneumopathies, pneumonia, lung tumors, pulmonary edema, thromboembolism, and bronchial diseases, especially in cats), severe abdominal effusion, hepatomegaly, gastric volvulus, peritonitis, and ketoacidosis diabetes. For 6 patients (1 dog and 5 cats), no cause of PB was identified. Therefore, although PB was strongly associated with pleural diseases, it cannot be considered as a specific sign in cats (Sp, 0.58). It was a more specific sign of pleural diseases in dogs (Sp, 0.83).
It seems that PB is principally initiated by the increased work of inspiratory muscles, resulting in forced inspiration. The inspiration aspirates abdominal viscera toward the chest.3,4 However, in the instance of, for example, a large obstruction of the extrathoracic airways, a forced inspiration can also be observed.20 This could explain why upper airway obstruction is described as a cause of PB in humans.16,17 Similarly, in the present study, 1 cat with a laryngeal mass and 1 dog with tracheal collapse had PB.
Paradoxical breathing appeared to be a sensitive sign associated with pleural diseases in the present study, especially in cats (Se, 0.90). However, 13 dogs and 9 cats had a pleural space disease without expressing PB (pleural effusion occurred in 11 dogs and 9 cats, and pneumothorax occurred in 2 dogs).
From a clinical standpoint, predictive values may be more relevant because clinicians are usually confronted with dyspneic animals without knowing whether they have pleural disease. In this study, PB did not appear useful to accurately predict the presence of pleural diseases with regard to the low PPV in dogs or cats. However, the absence of PB may be useful to exclude pleural diseases in dyspneic dogs and cats (NPV, 0.91 and 0.87, respectively). Furthermore, our results suggest that immediate thoracocentesis for diagnostic and treatment purposes, especially in an emergency situation, is not likely to be helpful in the management of a dog or cat that does not express PB, and other procedures should be immediately considered.
Signalment of dogs with pleural diseases in our study did not differ significantly from that of other reports, such as that of Mellanby et al2 regarding median age (7.0 years), dog weight (> 25 kg), and morphology (mesocephalic dogs in both). However there were more males (63%) in the previous report2 than in the present study (43%).
In cats, sample population in the present study was very similar to that of the study by Davies and Forrester21 with regard to median age (4 years), sex (53% males), and breed distribution (predominance of domestic shorthair cats, 87% in the present study vs 89% in the Davies and Forrester21 study). Age, sex, and breeds were not different for dyspneic animals with and without PB. Therefore, PB did not isolate a particular subset of the population.
In our study, signalment of animals with PB was close to that of animals affected with pleural diseases, but with some differences. For example, small-breed dogs were predominant among dogs with PB, and large-breed dogs were predominant among dogs with pleural diseases. This difference probably results from the fact that some dogs expressing PB did not have pleural diseases and that these were mainly small breed (18/27 [66%] weighed < 10 kg).
The present study had several limitations. The studied populations were selected on the basis of a diagnosis of dyspnea. This automatically excludes patients with a possible diagnosis of pleural diseases that did not develop dyspnea. Our results are therefore applicable only to dyspneic animals and not to all animals with pleural diseases. Also, the exclusion of animals without radiographs may have excluded those with an end-stage disease that died before having radiography performed. Lack of a standardized method of recording data combined with the retrospective nature of the study may have led to the misclassification of some animals' respiratory pattern. The finding that PB was more frequently identified at the National Veterinary School of Alfort than at the Toulouse school could be a direct consequence of this lack of standardization. This could have biased the OR between PB and pleural diseases. Nevertheless, the link identified between PB and pleural diseases in our study was very strong, with an adjusted OR of 12.6 in dogs and 14.1 in cats. According to our results, this suggests that a dyspneic dog or cat with PB was 12.6 or 14.1 times as likely, respectively, to be affected by a pleural disease as a dyspneic dog or cat with another breathing pattern. The CIs associated with these ORs were wide, but the lower limit of the CIs remains much higher than 1. As a comparison, although it refers to a causality link that is absolutely not the case in our study, the OR linking cigarette smoking to lung cancer in humans is approximately 13.22
ABBREVIATIONS
| CI | Confidence interval |
| NPV | Negative predictive value |
| PB | Paradoxical breathing |
| PPV | Positive predictive value |
| Se | Sensitivity |
| Sp | Specificity |
STATA, version 9.1, StataCorp LP, College Station, Tex.
References
- 1.↑
Padrid P. Canine and feline pleural disease. Vet Clin North Am Small Anim Pract 2000; 30:1295–1307.
- 2.↑
Mellanby RJ, Villiers E, Herrtage ME. Canine pleural and mediastinal effusions: a retrospective study of 81 cases. J Small Anim Pract 2002; 43:447–451.
- 3.
De Troyer A, Leduc D. Role of pleural pressure in the coupling between the intercostal muscles and the ribs. J Appl Physiol 2007; 102:2332–2337.
- 4.
De Troyer A, Brunko E, Leduc D, et al. Reflex inhibition of canine inspiratory intercostals by diaphragmatic tension receptors. J Physiol 1999; 514:255–263.
- 5.
al-Kaisy AA, Chan VW, Perlas A. Respiratory effects of low-dose bupivacaine interscalene block. Br J Anaesth 1999; 82:217–220.
- 6.
Kohyama J, Shiiki T, Shimohira M, et al. Asynchronous breathing during sleep. Arch Dis Child 2001; 84:174–177.
- 7.
Hendricks JC. Respiratory muscle fatigue and failure. In: King LK, ed. Textbook of respiratory disease in dogs and cats. St Louis: WB Saunders Co, 2004;61–65.
- 8.
Pereira MC, Mussi RF, Massucio RA, et al. Idiopathic bilateral diaphragmatic paresis. J Bras Pneumol 2006; 32:481–485.
- 9.↑
Hartley L, Kinali M, Knight R, et al. A congenital myopathy with diaphragmatic weakness not linked to the SMARD1 locus. Neuromuscul Disord 2007; 17:174–179.
- 10.↑
Piastra M, De Luca D, Zorzi G, et al. Noninvasive ventilation in large postoperative flail chest. Pediatr Blood Cancer 2008; 51:831–833.
- 11.
McKenzie DK, Butler JE, Gandevia SC. Respiratory muscle function and activation in chronic obstructive pulmonary disease. J Appl Physiol 2009; 107:621–629.
- 12.
Schilero GJ, Spungen AM, Bauman WA, et al. Pulmonary function and spinal cord injury. Respir Physiol Neurobiol 2009; 166:129–141.
- 13.
Matsumoto H, Nakayama T, Hamaguchi H, et al. Diaphragmatic paralysis in a patient with spinal cord infarction. Intern Med 2009; 48:1763–1766.
- 14.
Konagaya M, Yasuma F, Kuru S. Multiple system atrophy with paradoxical respiration. Respiration 2001; 68:527.
- 15.↑
Aliverti A, Quaranta M, Chakrabarti B, et al. Paradoxical movement of the lower ribcage at rest and during exercise in COPD patients. Eur Respir J 2009; 33:49–60.
- 16.
Resta O, Barbaro MP, Giliberti T, et al. Sleep related breathing disorders in adults with Down syndrome. Downs Syndr Res Pract 2003; 8:115–119.
- 17.
Yamakage M, Kamada Y, Toriyabe M, et al. Changes in respiratory pattern and arterial blood gases during sedation with propofol or midazolam in spinal anesthesia. J Clin Anesth 1999; 11:375–379.
- 18.↑
Anzai Y, Ohya T. A case of effective gastrostomy for severe abdominal distention due to breathing dysfunction of Rett's syndrome: a treatment of autonomic disorder. Brain Dev 2001; 23(suppl 1):S240–S241.
- 19.↑
Wang LM, Cherng JM, Wang JS. Improved lung function after thoracocentesis in patients with paradoxical movement of a hemidiaphragm secondary to a large pleural effusion. Respirology 2007; 12:719–723.
- 21.↑
Davies C, Forrester SD. Pleural effusion in cats: 82 cases (1987 to 1995). J Small Anim Pract 1996; 37:217–224.
- 22.↑
Barbone F, Bovenzi M, Cavallieri F, et al. Cigarette smoking and histologic type of lung cancer in men. Chest 1997; 112:1474–1479.
