REVIEW ARTICLE
Influenza Infection and Diabetes Mellitus
Rob J.A. Diepersloot, MD, PhD Karel P. Bouter, MD, PhD Joost B.L. Hoekstra, MD, PhD
Case for Annual Vaccination
Herein, epidemiological data on influenza pneumonia and mortality, results of clinical studies, and the outcome of influenza vaccination trials are reviewed. All excess mortality studies that specify for underlying disease list diabetes as one of the major risk factors. During influenza epidemics, death rates among patients with diabetes mellitus may increase by 5-15%. Diabetes mellitus is also mentioned as a risk factor in most clinical studies, making up 3-14% of the patients studied. Even in recent studies, diabetes mellitus is only preceded as a risk factor by cardiovascular disease and chronic pulmonary disorders. To what extent cardiovascular disease and old age contribute to the increased influenza mortality and morbidity in diabetic patients remains unclear. The influence of epidemic influenza on the incidence of diabetic acidosis in combination with an impaired immune response to both Staphylococcus aureus and the influenza virus suggests that diabetes mellitus itself is the main risk factor. It is concluded that all patients with diabetes mellitus should receive annual vaccinations and that, in official recommendations, patients with diabetes mellitus should be mentioned as a separate risk group. Wholevirus vaccines are preferred over subunit vaccines. Diabetes Care 13:876-82, 1990
Influenza, a viral agent discovered only 2 years before the death of Leonard Thompson, has since been incriminated as a cause of considerable morbidity and mortality in patients with diabetes mellitus. In more recent years, influenza morbidity and mortality have been mainly ascribed to underlying cardiovascular and chronic respiratory disease (3,4). The reasons for an increased mortality in patients suffering from these conditions are obvious. In patients with diabetes mellitus, however, the situation is more complicated. On the one hand, it can be argued that influenza mortality is mostly due to conditions like cardiovascular disease and renal impairment that are secondary to the diabetic state; on the other hand, diabetes by means of dysregulation and an impaired immune response may put patients at a high risk in the case of influenza infection. To elucidate to what extent various risk factors contribute to the excess morbidity and mortality in patients with diabetes mellitus, this review analyzes the epidemiological data on influenza pneumonia and mortality, the results of clinical studies, and the outcome of influenza vaccination trials.
EXCESS MORTALITY STUDIES
I
n 1935, Leonard Thompson, the first patient in the world to receive insulin, died in an oxygen tent at the age of 27 yr. The cause of death was a staphylococcal pneumonia complicating a respiratory infection, most likely influenza (1,2).
From the Department of Virology, Erasmus University, Rotterdam; and the Department of Internal Medicine, Diakonessen Hospital, Utrecht, Netherlands. Address correspondence and reprint requests to Rob J.A. Diepersloot, MD, PhD, Laboratory for Medical Microbiology, Herman Corterlaan 2, 5644 SW, Eindhoven, Netherlands. Received for publication 29 December 1988 and accepted in revised form 4 April 1990.
876
Since the original study of William Farr on the London influenza epidemic of 1847 (5), excess mortality figures have been the main tool in expressing the impact of influenza epidemics on public health. Not all studies express excess mortality by selected specified underlying disease (6,7). The studies that do, however, all list diabetes as one of the major risk factors (3,8-12). Polak (8), describing the 1957 epidemic in the Netherlands, ranks diabetes among other high-risk conditions, e.g., asthma, Parkinson's disease, tuberculosis, multiple sclerosis, scoliosis, and cardiac valvular leDIABETES CARE, VOL. 13, NO. 8, AUGUST 1990
R.J.A. DIEPERSLOOT, K.P. BOUTER, AND J.B.L HOEKSTRA
sions. Eickhoff et al. (9), studying the same epidemic in the United States, found an increased risk of death for patients suffering from cirrhosis, tuberculosis, rheumatic heart disease, asthma, chronic nephritis, and diabetes (9). Housworth and Langmuir (10) concluded that, during 1957-1966 (covering 7 influenza epidemics), excess mortality from tuberculosis, asthma, and chronic rheumatic heart disease was significant during the more intense influenza A epidemics but was either insignificant or barely significant during the milder influenza B epidemics. Arteriosclerotic heart disease was the only subclassification that showed significant excess during all epidemic periods, but excess deaths from diabetes were significant in six of the seven epidemics, including the influenza B epidemic of 1962. From the figures in this article, it can be calculated that death rates in patients with diabetes mellitus increased by —5-12% during epidemic periods (10). A similar increase in death rates (5-15%) among patients with diabetes mellitus in epidemic years was noted by Stocks (11) reporting on influenza mortality from 1921 to 1931. More accurate calculations on absolute and relative risks are presented by Barker and Mullooly (3) in their study on influenza deaths during the 1969 and 1973 influenza A (type H3N2) epidemic in Oregon. Relative risks in people >45 yrof age ranged from 39 for patients with one high-risk condition to 200 for patients with two or more high-risk conditions. Estimated death rates ranged from 2 deaths/100,000 among people aged 45-64 yr without chronic disease to 797 deaths/100,000 in people >65 yr with two or more high-risk conditions. The highest estimated rates involved people with cardiovascular disease in combination with either diabetes or chronic pulmonary disease (3). Cameron et al. (12) calculated a considerably lower relative mortality risk of 2.0 (range 0.4-14.8) for patients with diabetes mellitus in South Australia during 1969-1981. Their figures were based on death certificate data, which may lead to underestimation of actual mortality risks by ascribing mortality to other causes, cardiovascular diseases in particular (13,14). In a recent study on the effect of epidemic influenza on ketoacidosis, pneumonia, and death rates in patients with diabetes mellitus, Bouter et al. (15) calculated the relative risk for these patients to be hospitalized due to influenza to be 5.7-6.2 in epidemic years. From their data, they concluded that mortality in diabetic patients hospitalized due to pneumonia or ketoacidosis rose significantly from 14 to 25% in epidemic years (P < 0.05). Risk factors for influenza infection tend to accumulate in the elderly, and because most fatal cases of influenza occur in individuals aged >65 yr, it is difficult to assess the impact of each individual risk factor (7,16). Combinations of risk factors increase mortality risks exponentially (3,16). In the case of diabetes, cardiovascular disease is the most important risk factor that attributes to influenza-associated mortality. Barker and Mullooly (3) mention that, in their study, all patients with diabetes mellitus also had underlying cardiovascular disease. The combination of these risk factors bears a considerably DIABETES CARE, VOL. 13, NO. 8, AUGUST 1990
higher risk than cardiovascular disease alone. Polak (8), describing the 1957 epidemic in the Netherlands, also mentioned the occurrence of more than one risk factor in lethal influenza. For diabetes, they noted a combination with other risk factors in only 2 of 27 lethal cases (9). As demonstrated by Bouter et al. (15), influenza enhances mortality in the case of diabetic acidosis. Although diabetic acidosis is by definition restricted to diabetic individuals, mortality from this condition also is strongly influenced by cardiovascular disease (17). In conclusion, mortality in patients with diabetes mellitus increases during influenza epidemics. Mortality in these patients is especially high with regard to additional risk factors, in particular cardiovascular disease and age >65 yr.
CLINICAL STUDIES Clinical studies are not meant to provide epidemiological data but can give more detailed information on clinical features and, in some cases, disclose results of postmortem examinations. Results of clinical studies are shown in Table 1. Only studies with detailed information on the underlying disease are included. Patients with diabetes mellitus were reported in all but one study, making up 3-14% of the patients studied (13,18-26). Diabetes mellitus is not mentioned as one of the underlying diseases in the study by Winterbauer et al. (22). This may be explained by the small size (n = 11) of the patient group studied and the fact that only patients with viral pneumonia not complicated by secondary bacterial infection were included. The clinical and pathological findings of the patients presented in the reported studies fit well within the classification originally described by Hers et al. (27) and Louria et al. (28). They defined four basic clinical syndromes: 7) influenza virus alone, causing moderately
TABLE 1 Outcome of clinical studies on influenza infections
Infections Pneumonia (n) 24 91 79 108 11
Diabetic patients
Percent with diabetes
2 4 8 15 0
8 4-5
2 2
9
10 14 0
Publication (yr)
Ref.
1942 1959 1971 1971 1977
18 19 20 21
1950 1957 1959 1959 1981
23 24
22
Influenza deaths (n) 22 46 32 33 38
2 1 1
4 6 3 3
25 26 13
877
INFLUENZA INFECTION AND DIABETES MELLITUS
severe tracheitis and/or bronchiolitis; 2) a fulminating usually fatal pneumonia; 3) bacterial pneumonia that might either coexist with acute influenzal infection or 4) present as a postinfluenzal complication. Lethal viral pneumonia is especially noted in patients with rheumatic heart disease and patients with mitral stenosis (8,20,22-25). Martin et al. (25), however, report a case of fatal viral pneumonia in a 44-yr-old patient with diabetes mellitus. Further clinical information on diabetic patients is scarce. Stuart-Harris et al. (23) report diabetic coma in one patient and secondary staphylococcal infection in another diabetic patient. Both patients were >60 yr of age. The second patient with diabetes mellitus described by Martin et al. (25) (aged 30 yr) had postinfluenza nonstaphylococcal pneumonia. The two diabetic patients with type I (insulin-dependent) diabetes described by Finland et al. (18) were both male, 44 and 46 yr of age, and had staphylococcal pneumonia. Diabetic ketoacidosis is reported in one of seven fatal cases by Schwarzmann et al. (21). Apart from diabetic ketoacidosis, clinical data of patients with diabetes mellitus do not differ from those of other patients. Information on age, if available, shows that most of them were relatively young. Only one of the patients mentioned above was reported to have concomitant cardiovascular disease. Staphylococcus aureus is the main cause of secondary bacterial infection in most studies (18,23-26) and Streptococcus pneumoniae in other studies (19-21). Schwarzmann etal. (21) noted a sharp increase of staphylococcal pneumonia during the influenza epidemic compared with a nonepidemic period. Remarkably, in the same study, they reported a similar percentage of diabetes as an underlying disease in an epidemic (14%) and a nonepidemic (15%) period (21). Although the underlying high-risk conditions may vary over years (rheumatic heart disease and poliomyelitis mentioned in earlier studies and pregnancy only during major epidemics), diabetes is reported in all studies with remarkable consistency. In more recent studies, cardiovascular disease is the most important risk factor encountered in 20-40% of cases with influenza-associated pneumonia, followed by chronic pulmonary disorder (10-25%) and diabetes ranking third (3-14%; 13,20,21). Clinical studies, however, are not designed to obtain epidemiological data, so the figures given on underlying disease should be looked on with caution and cannot be regarded as an indication for relative morbidity or mortality risks.
5. AUREUS
As discussed above, secondary bacterial bronchopneumonia is one of the major complications in influenza infection (18,23-26). Hers et al. (29,31), Roberts (30), and Walker et al. (32) demonstrated that influenza virus can disrupt the respiratory epithelium extending to the
878
alveoli, and in this manner, give free access to the invading staphylococci (29-32). Several authors have reported antecedent staphylococcal skin lesions in patients with influenza and secondary staphylococcal pneumonia (31,33,34). Goslings et al. (35) could correlate 55% of 57 cases of secondary staphylococcal pneumonia to preceding staphylococcal skin infection in the patient or close relatives by phage typing. Overt lesions were the most common, with furunculosis in most cases, and folliculitis, pyodermia, and infected skin wounds were also evident. Staphylococcal skin infections are frequently reported in patients with diabetes mellitus, especially in poorly controlled patients and patients with foot ulcers (36). Moreover, phagocytosis and intracellular killing of 5. aureus have been demonstrated to be decreased in patients with diabetes mellitus (37-40). The impairment of phagocytosis and intracellular killing is most prominent in poorly controlled patients. As discussed later, influenza infection may provoke severe metabolic deterioration and, even in the case of a minor upper-respiratory infection, glucose levels may rise substantially (41). Usually this temporary loss of glucose control will go unnoticed. Although it seems rational to assume that the combination of increased carrier rate and impaired immune response to 5. aureus will put patients with diabetes at an increased risk during influenza epidemics, this hypothesis is based on circumstantial evidence only. There are no clinical or epidemiological data that link secondary staphylococcal pneumonia to antecedent staphylococcal skin infection in diabetic patients.
DIABETIC KETOACIDOSIS The incidence of ketoacidosis increases during winter months and is considered to be associated with resp-iratory infection (42,43). In addition to the patients mentioned above, ketoacidosis during influenza infection has been reported by several authors (44-46). Watkins et al. (46), studying diabetic ketoacidosis during an influenza epidemic, reported 29 cases over 8 wk. This was an exceptionally large number of cases because the annual number of patients admitted to their hospital with ketoacidosis each year was
Influenza Infection and Diabetes Mellitus
Rob J.A. Diepersloot, MD, PhD Karel P. Bouter, MD, PhD Joost B.L. Hoekstra, MD, PhD
Case for Annual Vaccination
Herein, epidemiological data on influenza pneumonia and mortality, results of clinical studies, and the outcome of influenza vaccination trials are reviewed. All excess mortality studies that specify for underlying disease list diabetes as one of the major risk factors. During influenza epidemics, death rates among patients with diabetes mellitus may increase by 5-15%. Diabetes mellitus is also mentioned as a risk factor in most clinical studies, making up 3-14% of the patients studied. Even in recent studies, diabetes mellitus is only preceded as a risk factor by cardiovascular disease and chronic pulmonary disorders. To what extent cardiovascular disease and old age contribute to the increased influenza mortality and morbidity in diabetic patients remains unclear. The influence of epidemic influenza on the incidence of diabetic acidosis in combination with an impaired immune response to both Staphylococcus aureus and the influenza virus suggests that diabetes mellitus itself is the main risk factor. It is concluded that all patients with diabetes mellitus should receive annual vaccinations and that, in official recommendations, patients with diabetes mellitus should be mentioned as a separate risk group. Wholevirus vaccines are preferred over subunit vaccines. Diabetes Care 13:876-82, 1990
Influenza, a viral agent discovered only 2 years before the death of Leonard Thompson, has since been incriminated as a cause of considerable morbidity and mortality in patients with diabetes mellitus. In more recent years, influenza morbidity and mortality have been mainly ascribed to underlying cardiovascular and chronic respiratory disease (3,4). The reasons for an increased mortality in patients suffering from these conditions are obvious. In patients with diabetes mellitus, however, the situation is more complicated. On the one hand, it can be argued that influenza mortality is mostly due to conditions like cardiovascular disease and renal impairment that are secondary to the diabetic state; on the other hand, diabetes by means of dysregulation and an impaired immune response may put patients at a high risk in the case of influenza infection. To elucidate to what extent various risk factors contribute to the excess morbidity and mortality in patients with diabetes mellitus, this review analyzes the epidemiological data on influenza pneumonia and mortality, the results of clinical studies, and the outcome of influenza vaccination trials.
EXCESS MORTALITY STUDIES
I
n 1935, Leonard Thompson, the first patient in the world to receive insulin, died in an oxygen tent at the age of 27 yr. The cause of death was a staphylococcal pneumonia complicating a respiratory infection, most likely influenza (1,2).
From the Department of Virology, Erasmus University, Rotterdam; and the Department of Internal Medicine, Diakonessen Hospital, Utrecht, Netherlands. Address correspondence and reprint requests to Rob J.A. Diepersloot, MD, PhD, Laboratory for Medical Microbiology, Herman Corterlaan 2, 5644 SW, Eindhoven, Netherlands. Received for publication 29 December 1988 and accepted in revised form 4 April 1990.
876
Since the original study of William Farr on the London influenza epidemic of 1847 (5), excess mortality figures have been the main tool in expressing the impact of influenza epidemics on public health. Not all studies express excess mortality by selected specified underlying disease (6,7). The studies that do, however, all list diabetes as one of the major risk factors (3,8-12). Polak (8), describing the 1957 epidemic in the Netherlands, ranks diabetes among other high-risk conditions, e.g., asthma, Parkinson's disease, tuberculosis, multiple sclerosis, scoliosis, and cardiac valvular leDIABETES CARE, VOL. 13, NO. 8, AUGUST 1990
R.J.A. DIEPERSLOOT, K.P. BOUTER, AND J.B.L HOEKSTRA
sions. Eickhoff et al. (9), studying the same epidemic in the United States, found an increased risk of death for patients suffering from cirrhosis, tuberculosis, rheumatic heart disease, asthma, chronic nephritis, and diabetes (9). Housworth and Langmuir (10) concluded that, during 1957-1966 (covering 7 influenza epidemics), excess mortality from tuberculosis, asthma, and chronic rheumatic heart disease was significant during the more intense influenza A epidemics but was either insignificant or barely significant during the milder influenza B epidemics. Arteriosclerotic heart disease was the only subclassification that showed significant excess during all epidemic periods, but excess deaths from diabetes were significant in six of the seven epidemics, including the influenza B epidemic of 1962. From the figures in this article, it can be calculated that death rates in patients with diabetes mellitus increased by —5-12% during epidemic periods (10). A similar increase in death rates (5-15%) among patients with diabetes mellitus in epidemic years was noted by Stocks (11) reporting on influenza mortality from 1921 to 1931. More accurate calculations on absolute and relative risks are presented by Barker and Mullooly (3) in their study on influenza deaths during the 1969 and 1973 influenza A (type H3N2) epidemic in Oregon. Relative risks in people >45 yrof age ranged from 39 for patients with one high-risk condition to 200 for patients with two or more high-risk conditions. Estimated death rates ranged from 2 deaths/100,000 among people aged 45-64 yr without chronic disease to 797 deaths/100,000 in people >65 yr with two or more high-risk conditions. The highest estimated rates involved people with cardiovascular disease in combination with either diabetes or chronic pulmonary disease (3). Cameron et al. (12) calculated a considerably lower relative mortality risk of 2.0 (range 0.4-14.8) for patients with diabetes mellitus in South Australia during 1969-1981. Their figures were based on death certificate data, which may lead to underestimation of actual mortality risks by ascribing mortality to other causes, cardiovascular diseases in particular (13,14). In a recent study on the effect of epidemic influenza on ketoacidosis, pneumonia, and death rates in patients with diabetes mellitus, Bouter et al. (15) calculated the relative risk for these patients to be hospitalized due to influenza to be 5.7-6.2 in epidemic years. From their data, they concluded that mortality in diabetic patients hospitalized due to pneumonia or ketoacidosis rose significantly from 14 to 25% in epidemic years (P < 0.05). Risk factors for influenza infection tend to accumulate in the elderly, and because most fatal cases of influenza occur in individuals aged >65 yr, it is difficult to assess the impact of each individual risk factor (7,16). Combinations of risk factors increase mortality risks exponentially (3,16). In the case of diabetes, cardiovascular disease is the most important risk factor that attributes to influenza-associated mortality. Barker and Mullooly (3) mention that, in their study, all patients with diabetes mellitus also had underlying cardiovascular disease. The combination of these risk factors bears a considerably DIABETES CARE, VOL. 13, NO. 8, AUGUST 1990
higher risk than cardiovascular disease alone. Polak (8), describing the 1957 epidemic in the Netherlands, also mentioned the occurrence of more than one risk factor in lethal influenza. For diabetes, they noted a combination with other risk factors in only 2 of 27 lethal cases (9). As demonstrated by Bouter et al. (15), influenza enhances mortality in the case of diabetic acidosis. Although diabetic acidosis is by definition restricted to diabetic individuals, mortality from this condition also is strongly influenced by cardiovascular disease (17). In conclusion, mortality in patients with diabetes mellitus increases during influenza epidemics. Mortality in these patients is especially high with regard to additional risk factors, in particular cardiovascular disease and age >65 yr.
CLINICAL STUDIES Clinical studies are not meant to provide epidemiological data but can give more detailed information on clinical features and, in some cases, disclose results of postmortem examinations. Results of clinical studies are shown in Table 1. Only studies with detailed information on the underlying disease are included. Patients with diabetes mellitus were reported in all but one study, making up 3-14% of the patients studied (13,18-26). Diabetes mellitus is not mentioned as one of the underlying diseases in the study by Winterbauer et al. (22). This may be explained by the small size (n = 11) of the patient group studied and the fact that only patients with viral pneumonia not complicated by secondary bacterial infection were included. The clinical and pathological findings of the patients presented in the reported studies fit well within the classification originally described by Hers et al. (27) and Louria et al. (28). They defined four basic clinical syndromes: 7) influenza virus alone, causing moderately
TABLE 1 Outcome of clinical studies on influenza infections
Infections Pneumonia (n) 24 91 79 108 11
Diabetic patients
Percent with diabetes
2 4 8 15 0
8 4-5
2 2
9
10 14 0
Publication (yr)
Ref.
1942 1959 1971 1971 1977
18 19 20 21
1950 1957 1959 1959 1981
23 24
22
Influenza deaths (n) 22 46 32 33 38
2 1 1
4 6 3 3
25 26 13
877
INFLUENZA INFECTION AND DIABETES MELLITUS
severe tracheitis and/or bronchiolitis; 2) a fulminating usually fatal pneumonia; 3) bacterial pneumonia that might either coexist with acute influenzal infection or 4) present as a postinfluenzal complication. Lethal viral pneumonia is especially noted in patients with rheumatic heart disease and patients with mitral stenosis (8,20,22-25). Martin et al. (25), however, report a case of fatal viral pneumonia in a 44-yr-old patient with diabetes mellitus. Further clinical information on diabetic patients is scarce. Stuart-Harris et al. (23) report diabetic coma in one patient and secondary staphylococcal infection in another diabetic patient. Both patients were >60 yr of age. The second patient with diabetes mellitus described by Martin et al. (25) (aged 30 yr) had postinfluenza nonstaphylococcal pneumonia. The two diabetic patients with type I (insulin-dependent) diabetes described by Finland et al. (18) were both male, 44 and 46 yr of age, and had staphylococcal pneumonia. Diabetic ketoacidosis is reported in one of seven fatal cases by Schwarzmann et al. (21). Apart from diabetic ketoacidosis, clinical data of patients with diabetes mellitus do not differ from those of other patients. Information on age, if available, shows that most of them were relatively young. Only one of the patients mentioned above was reported to have concomitant cardiovascular disease. Staphylococcus aureus is the main cause of secondary bacterial infection in most studies (18,23-26) and Streptococcus pneumoniae in other studies (19-21). Schwarzmann etal. (21) noted a sharp increase of staphylococcal pneumonia during the influenza epidemic compared with a nonepidemic period. Remarkably, in the same study, they reported a similar percentage of diabetes as an underlying disease in an epidemic (14%) and a nonepidemic (15%) period (21). Although the underlying high-risk conditions may vary over years (rheumatic heart disease and poliomyelitis mentioned in earlier studies and pregnancy only during major epidemics), diabetes is reported in all studies with remarkable consistency. In more recent studies, cardiovascular disease is the most important risk factor encountered in 20-40% of cases with influenza-associated pneumonia, followed by chronic pulmonary disorder (10-25%) and diabetes ranking third (3-14%; 13,20,21). Clinical studies, however, are not designed to obtain epidemiological data, so the figures given on underlying disease should be looked on with caution and cannot be regarded as an indication for relative morbidity or mortality risks.
5. AUREUS
As discussed above, secondary bacterial bronchopneumonia is one of the major complications in influenza infection (18,23-26). Hers et al. (29,31), Roberts (30), and Walker et al. (32) demonstrated that influenza virus can disrupt the respiratory epithelium extending to the
878
alveoli, and in this manner, give free access to the invading staphylococci (29-32). Several authors have reported antecedent staphylococcal skin lesions in patients with influenza and secondary staphylococcal pneumonia (31,33,34). Goslings et al. (35) could correlate 55% of 57 cases of secondary staphylococcal pneumonia to preceding staphylococcal skin infection in the patient or close relatives by phage typing. Overt lesions were the most common, with furunculosis in most cases, and folliculitis, pyodermia, and infected skin wounds were also evident. Staphylococcal skin infections are frequently reported in patients with diabetes mellitus, especially in poorly controlled patients and patients with foot ulcers (36). Moreover, phagocytosis and intracellular killing of 5. aureus have been demonstrated to be decreased in patients with diabetes mellitus (37-40). The impairment of phagocytosis and intracellular killing is most prominent in poorly controlled patients. As discussed later, influenza infection may provoke severe metabolic deterioration and, even in the case of a minor upper-respiratory infection, glucose levels may rise substantially (41). Usually this temporary loss of glucose control will go unnoticed. Although it seems rational to assume that the combination of increased carrier rate and impaired immune response to 5. aureus will put patients with diabetes at an increased risk during influenza epidemics, this hypothesis is based on circumstantial evidence only. There are no clinical or epidemiological data that link secondary staphylococcal pneumonia to antecedent staphylococcal skin infection in diabetic patients.
DIABETIC KETOACIDOSIS The incidence of ketoacidosis increases during winter months and is considered to be associated with resp-iratory infection (42,43). In addition to the patients mentioned above, ketoacidosis during influenza infection has been reported by several authors (44-46). Watkins et al. (46), studying diabetic ketoacidosis during an influenza epidemic, reported 29 cases over 8 wk. This was an exceptionally large number of cases because the annual number of patients admitted to their hospital with ketoacidosis each year was
