Hypoxemia in children with pneumonia and its clinical predictors
http://www.100md.com
《美国医学杂志》
1 Department of Pediatrics, Department of Community Medicine and Family Health, Kathmandu, Nepal
2 Department of Pediatrics, Institute of Medicine, Kathmandu, Nepal
Abstract
Objectives. To assess the prevalence of hypoxemia in children, 2 months to 5 years of age, with pneumonia and to identify its clinical predictors. Methods. Children between 2-60 months of age presenting with a complaint of cough or difficulty breathing were assessed. Hypoxemia was defined as an arterial oxygen saturation of <90% recorded by a portable pulse oximeter. Patients were categorized into groups: cough and cold, pneumonia, severe pneumonia and very severe pneumonia. Results. The prevalence of hypoxemia (SpO[2] of < 90%) in 150 children with pneumonia was 38.7%. Of them 100% of very severe pneumonia, 80% of severe and 17% of pneumonia patients were hypoxic. Number of infants with respiratory illness (p value = 0.03) and hypoxemia (Odds ratio = 2.21, 95% CI 1.03, 4.76) was significantly higher. Clinical predictors significantly associated with hypoxemia on univariate analysis were lethargy, grunting, nasal flaring, cyanosis, and complaint of inability to breastfeed/drink. Chest indrawing with 68.9% sensitivity and 82.6% specificity was the best predictor of hypoxemia. Conclusion. The prevalence and clinical predictors of hypoxemia identified validate the WHO classification of pneumonia based on severity. Age < 1 year in children with ARI is an important risk factor for hypoxemia.
Keywords: Pneumonia; Hypoxemia; Clinical predictors
Pneumonia is one of the common causes of morbidity and a significant cause of mortality in children under 5 years of age. The World Health Organization (WHO) estimates that acute respiratory infection (ARI), mostly in the form of pneumonia, is the leading cause of death in children under-five, killing over 2 million children annually.[1] Nepal, a developing country has an under five mortality rate of 91/1000 live births[2] and an infant mortality rate of 64/1000 live births.[2] Annual incidence of pneumonia in < 5 years is 90/1000, with 4.2/1000 children having severe pneumonia.[3] Hypoxemia in pneumonia has been shown to be a risk factor for death. Hypoxic children with pneumonia are five times more likely to die than those without in studies done in Kenya and Gambia.[4],[5] Such an association between hypoxemia and pneumonia suggests that its early detection and treatment are important aspects in the management of children with pneumonia. With the introduction of the pulse oximeter, a sufficiently accurate and non-invasive tool for measuring arterial oxygen saturation, assessment of hypoxemia in pneumonia has become possible. A pulse oximeter, however, is expensive and not suitable for routine use in a developing country like Nepal. To overcome this problem, clinical signs that best predict hypoxemia in pneumonia have been evaluated in earlier studies in other developing countries. Clinical features like cyanosis, tachypnea, grunting, head nodding, inability to cry, no spontaneous movement during clinical examination and chest retractions have been identified as the best clinical predictors of hypoxemia.[4],[5] The Integrated Management of Childhood Illnesses (IMCI) also makes use of clinical features to classify pneumonia into various categories based on severity.
At the Kanti Children's Hospital, although acute respiratory illness is the most frequent reason for admission and second most frequent cause of death, there is no provision to assess the degree of hypoxemia. Oxygen also does not seem to be given as frequently as one would think it should be used. Who are the children with pneumonia who really need oxygen as therapy Would certain clinical features help us to identify hypoxic children This study was done to find the prevalence of hypoxemia in children with pneumonia and identify its clinical predictors. It is based on the hypothesis that hypoxemia in pneumonia can be predicted clinically by assessing the various symptoms and signs associated with it.
Materials and Methods
A cross sectional study was conducted from 14th December 1999 to 18th July 2000 at the Kanti Children's Hospital in the Kathmandu valley that lies at an altitude of about 1336 meters above sea level. Children between 2 months to 5 years of age who presented to the Outpatient or the Emergency Department with a complaint of cough or difficulty breathing were assessed. Relevant history of the illness and examination was conducted according to a questionnaire prepared for the purpose of the study before the child received any form of treatment at the hospital. During the physical examination, arterial oxygen saturation was recorded using a portable, battery powered pulse oximeter (Mini SPO 2 T manufactured by the Criticare Systems, USA) with the sensor device placed over the finger (index or middle) or the big toe. A reading that was stable for at least 3 minutes was noted down. Hypoxemia was defined as an arterial oxygen saturation of <90% recorded by pulse oximetry. An arterial oxygen saturation of 90% generally corresponds to an arterial oxygen tension of 60-70mm Hg. This relation however is affected by factors such as temperature, pH, altitude and age. To overcome this problem, oxygen saturation by oximetry was recorded in children presenting with just cough and cold (assumed to be healthy children and therefore acting as the control group) to determine the normal range in all the children entered in the study.
Based on the clinical findings the patients were then categorized into four groups using the WHO guidelines [6]:
Patients with cough and cold
Patients with pneumonia
Patients with severe pneumonia
Patients with very severe pneumonia.
Patients were excluded from the study if:
A murmur was detected on auscultation of the heart indicating the presence of heart disease.
They could not be grouped into any of the categories mentioned above.
After completion of the examination, patients were either sent home with treatment or admitted to the hospital for further management.
Statistical analysis : This was done using the Epi Info Version 6 program in the computer. Clinical signs between hypoxemic and non-hypoxemic children were compared using the Chi square test, or by the Fisher's exact test if the expected frequencies were less than 5.
A p value of < or = 0.05 was considered significant. The sensitivity and specificity of each clinical sign in its ability to predict hypoxemia was also calculated.
Results
In the study conducted over a period of seven months, a total of 264 subjects were interviewed and examined. 14 children were excluded from the study, as they could not be classified into any category of illness as specified in the methodology.
The distribution of patients according to age and category of illness is shown in table1. Age range was 2 - 60 months with a median of 12 months (interquartile range 6 - 26). 123(49.2%) were females and 127(50.8%) were males. There was no difference in the number of males to females in all categories of illness (p value = 0.4). There were a significantly higher number of infants with respiratory illness (p value = 0.03). On assessment, none of the children with cough and cold had SpO 2 of < 90%. The mean oxygen saturation in this group was 96% (standard deviation of 2.123). In children with pneumonia, the SpO 2 recorded ranged from 50% to 99%. Prevalence of hypoxemia in children with pneumonia was 38.7%. Hypoxemia in children categorized according to the severity of their illness is given in table2.
On comparing oxygen saturation between the two age groups of patients with pneumonia Figure1, infants had a higher frequency of hypoxemia and the difference observed was statistically significant. (p value = 0.02, Odd's ratio 2.21, 95% CI 1.03, 4.76).
Symptoms and Signs Associated with Hypoxemia : table3 shows the distribution of prompted symptoms and clinical signs that were significantly associated with hypoxemia. Among symptoms, breathing difficulty, fast breathing and inability to feed/drink had significant association with hypoxemia. Only 2 children with very severe pneumonia had complaints of convulsions and that is probably the reason why the results did not reach statistical significance. Among the clinical signs, if the child was lethargic, had nasal flaring, central cyanosis, grunting, chest indrawing, tachypnea, tachycardia, fever or crepitations on auscultation, significant association with hypoxemia was found. Using WHO cut-off values for defining tachypnea: in the age group 2 -12 months, 36(87.8%) of the children with hypoxemia compared with 53(55.8%) without were tachypneic, p value of <0.001; in the 13-60 month olds, 17(100%) of the hypoxic children compared with 55(56.7%) without hypoxemia were tachypneic, p value of 0.001.
Sensitivity and specificity of clinical signs and symptoms : table3 also shows the sensitivity and specificity of signs and symptoms in hypoxemic and non-hypoxemic children that were significantly associated with hypoxemia. Chest indrawing had a sensitivity of 68.9% and specificity of 82.6%. Tachypnea in both age groups were found to be useful clinical signs with sensitivity of 90% and specificity of 43.6% in 2-12 month olds and sensitivity of 100% with specificity of 43.2% in 13-60 months of age. A complaint of difficulty breathing and fast breathing in a child by the caretakers, crepitations on auscultation of the chest and fever on examination were sensitive but non-specific indicators of hypoxemia. On the other hand, inability to feed, lethargy, nasal flaring, cyanosis and grunting were highly specific but relatively insensitive signs.
Discussion
In the study conducted at an altitude of approximately 1300m above sea level, prevalence of hypoxemia (SpO 2 of < 90%) in children 2-60 months old with pneumonia was 38.7%. Studies reporting prevalence of hypoxemia measured by pulse oximetry show wide variations and are not comparable because the cut-off values used to define hypoxemia; study population, setting and the altitude in which they were conducted differ.[4],[5],[7],[8],[9],[10],[11],[12]
In the present study, all the patients with very severe pneumonia, 80% with severe pneumonia and 18% with pneumonia were hypoxic. None of the children with cough and cold had hypoxemia. In a systematic review of published literature, Lozano also reports that the frequency of hypoxemia (pooled prevalence) is determined by the severity of the illness. While outpatient children and those with a clinical diagnosis of upper acute respiratory illness (ARI) had a low risk of hypoxemia (pooled estimate of 6-9%), the prevalence increased among hospitalized children (47%) and those with radiographically confirmed pneumonia (72%).[13]
Of the 250 enrolled children in this study there were a significantly higher number of infants and they were also noted to have significantly higher frequency of hypoxemia. Infants are vulnerable to acute respiratory infections because, not only do they have less mature immune systems[14] but are also unable to clear secretions. They also cannot verbally communicate their distress and this may predispose them to present with hypoxemia on arrival at the hospital. Other studies on hypoxemia in children with ARI have not assessed the role of age as a risk factor.[13] On the other hand, age below 12 months has been identified as a predictor of mortality in children with pneumonia.[15],[16] Although these studies do not prove that the children who were at risk of dying were hypoxemic, other studies have found an association between hypoxemia and mortality in children with pneumonia .[4],[5]
Symptoms reported by caretakers of difficulty breathing, fast breathing and inability to feed/drink was significantly associated with hypoxemia in the present study. Severely reduced feeding in studies by Usen[5] and Weber [10] and history of difficult respiration in infants by Onyango[4] was significantly associated with hypoxemia.
The signs significantly associated with hypoxemia in the study were lethargy, nasal flaring, central cyanosis, grunting, chest indrawing, crepitations and tachypnea. A child was called lethargic if he/she was drowsy and not showing interest in what was happening. Impaired rousability[5], unresponsiveness[4] and noted as being drowsy[12] in other studies had good correlation with hypoxemia. In these studies grunting was also significantly associated with hypoxemia. [4],[5],[12] Nasal flaring and cyanosis was significantly associated with hypoxemia in other studies.[5],[9],[12] Presence of chest indrawing in children with pneumonia is used to categorize pneumonia as severe and requiring admission.[6] This sign was also significantly associated with hypoxemia in the studies done in Kenya[4] and Delhi.[11] Other studies showing significant association between crepitations and hypoxemia were those by Onyango[4], Usen[5], Smyth[10] and Kabra.[11]
Tachypnea in the present study was defined using cut-off values based on age according to WHO guidelines. [6] Higher respiratory rates used in studies by Onyango[4] (> or = 60/min and 70/min), Usen[5] (>60/min, 70/min and 90/min) and Smyth[10] (>70/min) in infants were found to be significantly associated with hypoxemia. In order to compare findings with these studies a respiratory rate of > or = 70/min was analyzed and found to be significantly associated with hypoxemia.
Chest indrawing with a sensitivity of 68.9% and specificity of 82.6% was the best clinical predictor of hypoxemia in this study. In studies done by Reuland[7] and Weber[9], sensitivity/specificity of chest indrawing as a predictor of hypoxemia in 2-60 month old children was 35/94 and 49/60 respectively. Similarly in the study by Kabra [11], chest indrawing had a sensitivity of 35.7% and specificity 86.4% in children < 5 years of age. Although the findings differ, all studies are able to show that chest indrawing is not uniformly sensitive but a fairly specific sign in the prediction of hypoxemia. Absence of this sign is likely to miss only a small percentage of patients with pneumonia who are also hypoxemic.
Tachypnea in patients 2-12 months of age (respiratory rate of > or = 50/min) predicted hypoxemia with a sensitivity of 90% and specificity of 43.6% in the present study. In Lozano's study similar respiratory rates in infants had a sensitivity of 76% and specificity of 71%.[8] In Smyth's study in the same age group of children higher respiratory rate of >70/min had 63% sensitivity and 89% specificity.[10] In Onyango's study sensitivity of 86% and specificity of 56% with respiratory rate of > or = 60/min changed to 51% sensitivity and 83% specificity when respiratory rate was > or = 70/min in infants. [4] Similarly in the study by Kabra, increasing the respiratory cut off by 10/minute lead to a decline in sensitivity from 82.1% to 53.6% and increment in specificity from 51.8% to 77.8%.[11] The present study was also able to demonstrate a decrease in sensitivity (42%) and rise in specificity (73%) when analysis was done using a respiratory rate of >/= 70/min in infants. An elevated respiratory rate in a sick child with pneumonia could result from the metabolic acidosis secondary to the dehydration from fever, panting and inability to drink as well as decreased peripheral perfusion. This would limit the usefulness of increasing respiratory rate in assessing the degree of hypoxia.[17] In the age group of 13-60 months, a respiratory rate of > or =40/min was 100% sensitive and 43.2% specific in its ability to predict hypoxemia. While Lozano reported 73% sensitivity and 61% specificity using the same cut-off values, [8] other studies have used higher respiratory rates to calculate the sensitivity and specificity. Further analysis using higher respiratory rates to compare findings was not done in this sub-group.
This study shows a higher frequency of hypoxemia in children with increasing severity of pneumonia. Clinical predictors significantly associated with hypoxemia on univariate analysis (lethargy, grunting, nasal flaring, cyanosis, and complaint of inability to breastfeed/drink) are those used for the recognition of very severe pneumonia in 2-60 month old children by WHO.[6] The study therefore validates WHO criteria for the recognition of children with severe and very severe pneumonia. It also supports the findings by Lozano that hypoxemia is more frequent in those with increasing severity of ARI. [13] Age < or = 12 months as a risk factor for hypoxemia in children with ARI, which has not been reported in other studies is an important finding of this study. These are the strengths of this study.
This study and a review of the literature of similar studies show that no single clinical sign can predict hypoxemia with both high sensitivity and specificity. Other studies have used combination models to improve the predictive value of clinical signs. While constructing models, addition of signs improves sensitivity but decreases specificity and this would be of benefit in settings where oxygen is freely available.[18] In a developing country like Nepal, with scarce resources the wastage of oxygen needs to be minimized. In this study, a combination model using signs for the prediction of hypoxemia in settings with limited supply of oxygen was not constructed. This is a major limitation of this study.
Conclusion
The increasing frequency of hypoxemia in children with more severe illness and the clinical predictors identified in the study validate the WHO classification of pneumonia based on severity. In Nepal where pulse oximeters may not always be available, simple clinical signs can still be used to identify hypoxemia, classify pneumonia as severe to very severe and administer oxygen. Children below 1 year of age with ARI are more likely to present with hypoxemia according to our findings. This subgroup of children should either be admitted for closer observation and frequent monitoring or followed up more vigorously. In the absence of pulse oximeters, they should receive oxygen earlier if they fail to respond to conservative management.
References
1. Reducing mortality from major killers of children. WHO Fact sheet No. 178, revised September 1998
2. Nepal Demographic and Health Survey 2001 , Family Health Division, Department of Health Services. Preliminary Report. September 2001.
3. Nepal Ministry of Health , Department of Health Services Annual Report 2000 - 2001.
4. Onyango FE, Steinhoff MC et al. Hypoxemia in young Kenyan children with acute lower respiratory tract infection. BMJ March 1993; 306 : 612-614.
5. Usen S, Weber M, Mulholland K et al. Clinical predictors of hypoxemia in Gambian children. BMJ Jan 1999; 318 : 86-91.
6. Management of the child with a serious infection or severe malnutrition. Guidelines For Care At The First Referral Level In Developing Countries . World Health Organization, 2000.
7. Reuland DS, Steinhoff MC, Gilman RH et al. Prevalence and prediction of hypoxemia in children with respiratory infection in the Peruvian Andes. J Pediatr Dec 1991; 119(6) : 900-906.
8. Lozano JM, Steinhoff MC, Ruiz JG et al. Clinical predictors of acute radiological pneumonia and hypoxemia at high altitude. Arch Dis Child Oct 1994;71(4) : 323-327.
9. Weber MW, Usen S, Palmer A et al. Predictors of hypoxemia in hospital admissions with acute lower respiratory tract infection in a developing country. Arch Dis Child April 1997; 76 : 310-314.
10. Smyth A, Carty H, Hart CA. Clinical predictors of hypoxemia in children with pneumonia. Ann Trop Pediatr Mar 1998; 18(1): 31-40.
11. Kabra SK, Lodha R et al. Can clinical symptoms and signs accurately predict the prevalence of hypoxemia in children with acute lower respiratory infections Indian Pediatrics 2004; 41 : 129-135.
12. Laman M, Ripa P, Vince J et al. Can clinical signs predict hypoxemia in Papua New Guinean children with moderate and severe pneumonia. Ann Trop Paediatr 2005; 25 : 31-40.
13. J.M. Lozano Epidemiology of hypoxemia in children with acute lower respiratory tract infection. Int J Tuberc Lung Dis 2001; 5(6): 496-504.
14. Regelmann WE, Hill HR, Cates KL, Quie PG. Immunology of the newborn. In: Feigin RD, Cherry JD, eds. Textbook of Pediatric Infectious Diseases. 3rd ed. Philadelphia: WB Saunders Company, Harcourt Brace Jovanovich Inc, 1992 876-887.
15. Sehgal V, Sethi GR, Sachdev HP et al. Predictors of mortality in subjects hospitalized with acute lower respiratory tract infections. Indian Pediatr Mar 1997; 34(3) : 213-219.
16. Banajeh SM. Outcome for children under 5 years hospitalized with severe acute lower respiratory tract infections in Yemen: a 5-year experience. J Trop Pediatr Dec 1998; 44(6): 343-346.
17. Dyke T, Brown N. Hypoxia in pneumonia: better detection and more oxygen needed in developing countries. BMJ Jan 1994; 308 : 119-120.
18. Usen S, Weber M. Clinical signs of hypoxemia in children with acute lower respiratory infection: indicators of oxygen therapy. Int J Tuberc Lung Dis 2001; 5(6): 505-510.(Basnet Sudha, Adhikari Ra)
2 Department of Pediatrics, Institute of Medicine, Kathmandu, Nepal
Abstract
Objectives. To assess the prevalence of hypoxemia in children, 2 months to 5 years of age, with pneumonia and to identify its clinical predictors. Methods. Children between 2-60 months of age presenting with a complaint of cough or difficulty breathing were assessed. Hypoxemia was defined as an arterial oxygen saturation of <90% recorded by a portable pulse oximeter. Patients were categorized into groups: cough and cold, pneumonia, severe pneumonia and very severe pneumonia. Results. The prevalence of hypoxemia (SpO[2] of < 90%) in 150 children with pneumonia was 38.7%. Of them 100% of very severe pneumonia, 80% of severe and 17% of pneumonia patients were hypoxic. Number of infants with respiratory illness (p value = 0.03) and hypoxemia (Odds ratio = 2.21, 95% CI 1.03, 4.76) was significantly higher. Clinical predictors significantly associated with hypoxemia on univariate analysis were lethargy, grunting, nasal flaring, cyanosis, and complaint of inability to breastfeed/drink. Chest indrawing with 68.9% sensitivity and 82.6% specificity was the best predictor of hypoxemia. Conclusion. The prevalence and clinical predictors of hypoxemia identified validate the WHO classification of pneumonia based on severity. Age < 1 year in children with ARI is an important risk factor for hypoxemia.
Keywords: Pneumonia; Hypoxemia; Clinical predictors
Pneumonia is one of the common causes of morbidity and a significant cause of mortality in children under 5 years of age. The World Health Organization (WHO) estimates that acute respiratory infection (ARI), mostly in the form of pneumonia, is the leading cause of death in children under-five, killing over 2 million children annually.[1] Nepal, a developing country has an under five mortality rate of 91/1000 live births[2] and an infant mortality rate of 64/1000 live births.[2] Annual incidence of pneumonia in < 5 years is 90/1000, with 4.2/1000 children having severe pneumonia.[3] Hypoxemia in pneumonia has been shown to be a risk factor for death. Hypoxic children with pneumonia are five times more likely to die than those without in studies done in Kenya and Gambia.[4],[5] Such an association between hypoxemia and pneumonia suggests that its early detection and treatment are important aspects in the management of children with pneumonia. With the introduction of the pulse oximeter, a sufficiently accurate and non-invasive tool for measuring arterial oxygen saturation, assessment of hypoxemia in pneumonia has become possible. A pulse oximeter, however, is expensive and not suitable for routine use in a developing country like Nepal. To overcome this problem, clinical signs that best predict hypoxemia in pneumonia have been evaluated in earlier studies in other developing countries. Clinical features like cyanosis, tachypnea, grunting, head nodding, inability to cry, no spontaneous movement during clinical examination and chest retractions have been identified as the best clinical predictors of hypoxemia.[4],[5] The Integrated Management of Childhood Illnesses (IMCI) also makes use of clinical features to classify pneumonia into various categories based on severity.
At the Kanti Children's Hospital, although acute respiratory illness is the most frequent reason for admission and second most frequent cause of death, there is no provision to assess the degree of hypoxemia. Oxygen also does not seem to be given as frequently as one would think it should be used. Who are the children with pneumonia who really need oxygen as therapy Would certain clinical features help us to identify hypoxic children This study was done to find the prevalence of hypoxemia in children with pneumonia and identify its clinical predictors. It is based on the hypothesis that hypoxemia in pneumonia can be predicted clinically by assessing the various symptoms and signs associated with it.
Materials and Methods
A cross sectional study was conducted from 14th December 1999 to 18th July 2000 at the Kanti Children's Hospital in the Kathmandu valley that lies at an altitude of about 1336 meters above sea level. Children between 2 months to 5 years of age who presented to the Outpatient or the Emergency Department with a complaint of cough or difficulty breathing were assessed. Relevant history of the illness and examination was conducted according to a questionnaire prepared for the purpose of the study before the child received any form of treatment at the hospital. During the physical examination, arterial oxygen saturation was recorded using a portable, battery powered pulse oximeter (Mini SPO 2 T manufactured by the Criticare Systems, USA) with the sensor device placed over the finger (index or middle) or the big toe. A reading that was stable for at least 3 minutes was noted down. Hypoxemia was defined as an arterial oxygen saturation of <90% recorded by pulse oximetry. An arterial oxygen saturation of 90% generally corresponds to an arterial oxygen tension of 60-70mm Hg. This relation however is affected by factors such as temperature, pH, altitude and age. To overcome this problem, oxygen saturation by oximetry was recorded in children presenting with just cough and cold (assumed to be healthy children and therefore acting as the control group) to determine the normal range in all the children entered in the study.
Based on the clinical findings the patients were then categorized into four groups using the WHO guidelines [6]:
Patients with cough and cold
Patients with pneumonia
Patients with severe pneumonia
Patients with very severe pneumonia.
Patients were excluded from the study if:
A murmur was detected on auscultation of the heart indicating the presence of heart disease.
They could not be grouped into any of the categories mentioned above.
After completion of the examination, patients were either sent home with treatment or admitted to the hospital for further management.
Statistical analysis : This was done using the Epi Info Version 6 program in the computer. Clinical signs between hypoxemic and non-hypoxemic children were compared using the Chi square test, or by the Fisher's exact test if the expected frequencies were less than 5.
A p value of < or = 0.05 was considered significant. The sensitivity and specificity of each clinical sign in its ability to predict hypoxemia was also calculated.
Results
In the study conducted over a period of seven months, a total of 264 subjects were interviewed and examined. 14 children were excluded from the study, as they could not be classified into any category of illness as specified in the methodology.
The distribution of patients according to age and category of illness is shown in table1. Age range was 2 - 60 months with a median of 12 months (interquartile range 6 - 26). 123(49.2%) were females and 127(50.8%) were males. There was no difference in the number of males to females in all categories of illness (p value = 0.4). There were a significantly higher number of infants with respiratory illness (p value = 0.03). On assessment, none of the children with cough and cold had SpO 2 of < 90%. The mean oxygen saturation in this group was 96% (standard deviation of 2.123). In children with pneumonia, the SpO 2 recorded ranged from 50% to 99%. Prevalence of hypoxemia in children with pneumonia was 38.7%. Hypoxemia in children categorized according to the severity of their illness is given in table2.
On comparing oxygen saturation between the two age groups of patients with pneumonia Figure1, infants had a higher frequency of hypoxemia and the difference observed was statistically significant. (p value = 0.02, Odd's ratio 2.21, 95% CI 1.03, 4.76).
Symptoms and Signs Associated with Hypoxemia : table3 shows the distribution of prompted symptoms and clinical signs that were significantly associated with hypoxemia. Among symptoms, breathing difficulty, fast breathing and inability to feed/drink had significant association with hypoxemia. Only 2 children with very severe pneumonia had complaints of convulsions and that is probably the reason why the results did not reach statistical significance. Among the clinical signs, if the child was lethargic, had nasal flaring, central cyanosis, grunting, chest indrawing, tachypnea, tachycardia, fever or crepitations on auscultation, significant association with hypoxemia was found. Using WHO cut-off values for defining tachypnea: in the age group 2 -12 months, 36(87.8%) of the children with hypoxemia compared with 53(55.8%) without were tachypneic, p value of <0.001; in the 13-60 month olds, 17(100%) of the hypoxic children compared with 55(56.7%) without hypoxemia were tachypneic, p value of 0.001.
Sensitivity and specificity of clinical signs and symptoms : table3 also shows the sensitivity and specificity of signs and symptoms in hypoxemic and non-hypoxemic children that were significantly associated with hypoxemia. Chest indrawing had a sensitivity of 68.9% and specificity of 82.6%. Tachypnea in both age groups were found to be useful clinical signs with sensitivity of 90% and specificity of 43.6% in 2-12 month olds and sensitivity of 100% with specificity of 43.2% in 13-60 months of age. A complaint of difficulty breathing and fast breathing in a child by the caretakers, crepitations on auscultation of the chest and fever on examination were sensitive but non-specific indicators of hypoxemia. On the other hand, inability to feed, lethargy, nasal flaring, cyanosis and grunting were highly specific but relatively insensitive signs.
Discussion
In the study conducted at an altitude of approximately 1300m above sea level, prevalence of hypoxemia (SpO 2 of < 90%) in children 2-60 months old with pneumonia was 38.7%. Studies reporting prevalence of hypoxemia measured by pulse oximetry show wide variations and are not comparable because the cut-off values used to define hypoxemia; study population, setting and the altitude in which they were conducted differ.[4],[5],[7],[8],[9],[10],[11],[12]
In the present study, all the patients with very severe pneumonia, 80% with severe pneumonia and 18% with pneumonia were hypoxic. None of the children with cough and cold had hypoxemia. In a systematic review of published literature, Lozano also reports that the frequency of hypoxemia (pooled prevalence) is determined by the severity of the illness. While outpatient children and those with a clinical diagnosis of upper acute respiratory illness (ARI) had a low risk of hypoxemia (pooled estimate of 6-9%), the prevalence increased among hospitalized children (47%) and those with radiographically confirmed pneumonia (72%).[13]
Of the 250 enrolled children in this study there were a significantly higher number of infants and they were also noted to have significantly higher frequency of hypoxemia. Infants are vulnerable to acute respiratory infections because, not only do they have less mature immune systems[14] but are also unable to clear secretions. They also cannot verbally communicate their distress and this may predispose them to present with hypoxemia on arrival at the hospital. Other studies on hypoxemia in children with ARI have not assessed the role of age as a risk factor.[13] On the other hand, age below 12 months has been identified as a predictor of mortality in children with pneumonia.[15],[16] Although these studies do not prove that the children who were at risk of dying were hypoxemic, other studies have found an association between hypoxemia and mortality in children with pneumonia .[4],[5]
Symptoms reported by caretakers of difficulty breathing, fast breathing and inability to feed/drink was significantly associated with hypoxemia in the present study. Severely reduced feeding in studies by Usen[5] and Weber [10] and history of difficult respiration in infants by Onyango[4] was significantly associated with hypoxemia.
The signs significantly associated with hypoxemia in the study were lethargy, nasal flaring, central cyanosis, grunting, chest indrawing, crepitations and tachypnea. A child was called lethargic if he/she was drowsy and not showing interest in what was happening. Impaired rousability[5], unresponsiveness[4] and noted as being drowsy[12] in other studies had good correlation with hypoxemia. In these studies grunting was also significantly associated with hypoxemia. [4],[5],[12] Nasal flaring and cyanosis was significantly associated with hypoxemia in other studies.[5],[9],[12] Presence of chest indrawing in children with pneumonia is used to categorize pneumonia as severe and requiring admission.[6] This sign was also significantly associated with hypoxemia in the studies done in Kenya[4] and Delhi.[11] Other studies showing significant association between crepitations and hypoxemia were those by Onyango[4], Usen[5], Smyth[10] and Kabra.[11]
Tachypnea in the present study was defined using cut-off values based on age according to WHO guidelines. [6] Higher respiratory rates used in studies by Onyango[4] (> or = 60/min and 70/min), Usen[5] (>60/min, 70/min and 90/min) and Smyth[10] (>70/min) in infants were found to be significantly associated with hypoxemia. In order to compare findings with these studies a respiratory rate of > or = 70/min was analyzed and found to be significantly associated with hypoxemia.
Chest indrawing with a sensitivity of 68.9% and specificity of 82.6% was the best clinical predictor of hypoxemia in this study. In studies done by Reuland[7] and Weber[9], sensitivity/specificity of chest indrawing as a predictor of hypoxemia in 2-60 month old children was 35/94 and 49/60 respectively. Similarly in the study by Kabra [11], chest indrawing had a sensitivity of 35.7% and specificity 86.4% in children < 5 years of age. Although the findings differ, all studies are able to show that chest indrawing is not uniformly sensitive but a fairly specific sign in the prediction of hypoxemia. Absence of this sign is likely to miss only a small percentage of patients with pneumonia who are also hypoxemic.
Tachypnea in patients 2-12 months of age (respiratory rate of > or = 50/min) predicted hypoxemia with a sensitivity of 90% and specificity of 43.6% in the present study. In Lozano's study similar respiratory rates in infants had a sensitivity of 76% and specificity of 71%.[8] In Smyth's study in the same age group of children higher respiratory rate of >70/min had 63% sensitivity and 89% specificity.[10] In Onyango's study sensitivity of 86% and specificity of 56% with respiratory rate of > or = 60/min changed to 51% sensitivity and 83% specificity when respiratory rate was > or = 70/min in infants. [4] Similarly in the study by Kabra, increasing the respiratory cut off by 10/minute lead to a decline in sensitivity from 82.1% to 53.6% and increment in specificity from 51.8% to 77.8%.[11] The present study was also able to demonstrate a decrease in sensitivity (42%) and rise in specificity (73%) when analysis was done using a respiratory rate of >/= 70/min in infants. An elevated respiratory rate in a sick child with pneumonia could result from the metabolic acidosis secondary to the dehydration from fever, panting and inability to drink as well as decreased peripheral perfusion. This would limit the usefulness of increasing respiratory rate in assessing the degree of hypoxia.[17] In the age group of 13-60 months, a respiratory rate of > or =40/min was 100% sensitive and 43.2% specific in its ability to predict hypoxemia. While Lozano reported 73% sensitivity and 61% specificity using the same cut-off values, [8] other studies have used higher respiratory rates to calculate the sensitivity and specificity. Further analysis using higher respiratory rates to compare findings was not done in this sub-group.
This study shows a higher frequency of hypoxemia in children with increasing severity of pneumonia. Clinical predictors significantly associated with hypoxemia on univariate analysis (lethargy, grunting, nasal flaring, cyanosis, and complaint of inability to breastfeed/drink) are those used for the recognition of very severe pneumonia in 2-60 month old children by WHO.[6] The study therefore validates WHO criteria for the recognition of children with severe and very severe pneumonia. It also supports the findings by Lozano that hypoxemia is more frequent in those with increasing severity of ARI. [13] Age < or = 12 months as a risk factor for hypoxemia in children with ARI, which has not been reported in other studies is an important finding of this study. These are the strengths of this study.
This study and a review of the literature of similar studies show that no single clinical sign can predict hypoxemia with both high sensitivity and specificity. Other studies have used combination models to improve the predictive value of clinical signs. While constructing models, addition of signs improves sensitivity but decreases specificity and this would be of benefit in settings where oxygen is freely available.[18] In a developing country like Nepal, with scarce resources the wastage of oxygen needs to be minimized. In this study, a combination model using signs for the prediction of hypoxemia in settings with limited supply of oxygen was not constructed. This is a major limitation of this study.
Conclusion
The increasing frequency of hypoxemia in children with more severe illness and the clinical predictors identified in the study validate the WHO classification of pneumonia based on severity. In Nepal where pulse oximeters may not always be available, simple clinical signs can still be used to identify hypoxemia, classify pneumonia as severe to very severe and administer oxygen. Children below 1 year of age with ARI are more likely to present with hypoxemia according to our findings. This subgroup of children should either be admitted for closer observation and frequent monitoring or followed up more vigorously. In the absence of pulse oximeters, they should receive oxygen earlier if they fail to respond to conservative management.
References
1. Reducing mortality from major killers of children. WHO Fact sheet No. 178, revised September 1998
2. Nepal Demographic and Health Survey 2001 , Family Health Division, Department of Health Services. Preliminary Report. September 2001.
3. Nepal Ministry of Health , Department of Health Services Annual Report 2000 - 2001.
4. Onyango FE, Steinhoff MC et al. Hypoxemia in young Kenyan children with acute lower respiratory tract infection. BMJ March 1993; 306 : 612-614.
5. Usen S, Weber M, Mulholland K et al. Clinical predictors of hypoxemia in Gambian children. BMJ Jan 1999; 318 : 86-91.
6. Management of the child with a serious infection or severe malnutrition. Guidelines For Care At The First Referral Level In Developing Countries . World Health Organization, 2000.
7. Reuland DS, Steinhoff MC, Gilman RH et al. Prevalence and prediction of hypoxemia in children with respiratory infection in the Peruvian Andes. J Pediatr Dec 1991; 119(6) : 900-906.
8. Lozano JM, Steinhoff MC, Ruiz JG et al. Clinical predictors of acute radiological pneumonia and hypoxemia at high altitude. Arch Dis Child Oct 1994;71(4) : 323-327.
9. Weber MW, Usen S, Palmer A et al. Predictors of hypoxemia in hospital admissions with acute lower respiratory tract infection in a developing country. Arch Dis Child April 1997; 76 : 310-314.
10. Smyth A, Carty H, Hart CA. Clinical predictors of hypoxemia in children with pneumonia. Ann Trop Pediatr Mar 1998; 18(1): 31-40.
11. Kabra SK, Lodha R et al. Can clinical symptoms and signs accurately predict the prevalence of hypoxemia in children with acute lower respiratory infections Indian Pediatrics 2004; 41 : 129-135.
12. Laman M, Ripa P, Vince J et al. Can clinical signs predict hypoxemia in Papua New Guinean children with moderate and severe pneumonia. Ann Trop Paediatr 2005; 25 : 31-40.
13. J.M. Lozano Epidemiology of hypoxemia in children with acute lower respiratory tract infection. Int J Tuberc Lung Dis 2001; 5(6): 496-504.
14. Regelmann WE, Hill HR, Cates KL, Quie PG. Immunology of the newborn. In: Feigin RD, Cherry JD, eds. Textbook of Pediatric Infectious Diseases. 3rd ed. Philadelphia: WB Saunders Company, Harcourt Brace Jovanovich Inc, 1992 876-887.
15. Sehgal V, Sethi GR, Sachdev HP et al. Predictors of mortality in subjects hospitalized with acute lower respiratory tract infections. Indian Pediatr Mar 1997; 34(3) : 213-219.
16. Banajeh SM. Outcome for children under 5 years hospitalized with severe acute lower respiratory tract infections in Yemen: a 5-year experience. J Trop Pediatr Dec 1998; 44(6): 343-346.
17. Dyke T, Brown N. Hypoxia in pneumonia: better detection and more oxygen needed in developing countries. BMJ Jan 1994; 308 : 119-120.
18. Usen S, Weber M. Clinical signs of hypoxemia in children with acute lower respiratory infection: indicators of oxygen therapy. Int J Tuberc Lung Dis 2001; 5(6): 505-510.(Basnet Sudha, Adhikari Ra)