Page Not Found | University of Alberta
Case series / outcomes of hospitalizations for swine-origin influenza A (H1N1) 2009
Perez-Padilla, 2009
- Case series of the first 18 persons with pneumonia and laboratory-confirmed S-OIV hospitalized at the INER hospital in Mexico. Retrospective chart review, covering March 24 to April 24, 2009.
- Most patients were young to middle-aged and previously healthy.
- 8/18 patients had pre-existing medical conditions. Non-type 1 diabetes was found in 3 patients (3/18 = 17%).
- Respiratory distress requiring intubation and mechanical ventilation in 10 patients during the first 24 hours, and in an additional 2 patients.
- Mortality | hospitalization = 7/18 (39%) patients died after a mean of 14 days of illness onset and after mean 9 days LOS.
- Mortality | ventilation = 7/12 (58%).
- Influenza-like illness that progressed during a period of 5 to 7 days, developed into pneumonia, and had findings of ARDS or acute lung injury on admission.
Novel Swine-Origin Influenza A (H1N1) Virus Investigation Team, 2009
- Case series of the first 642 confirmed cases of human infection in the US.
- April 15- May5, 2009
- Fever (94%), cough (92%), and sore throat (66%) most common symptoms. 25% of patients had diarrhea, and 25% had vomiting.
- 40% of patients between ages of 10 to 18 years, only 5% aged 51 years or older.
- 399 patients with known hospitalization status.
- Hospitalization | case = 36/399 (9%).
- 22 hospitalized patients with known comorbidity / outcome status.
- Chronic medical conditions in 9/22 patients (41%). Diabetes not mentioned.
- ICU | hospitalization = 8/22 (36%).
- Ventilation | hospitalization = 4/22 (18%).
- As of May 5th – Death | hospitalization = 2/22, in a 22-month-old child with neonatal myasthenia gravis and in a 33-year-old pregnant woman.
- Pneumonia admissions appear to be primary viral, no mention of secondary bacterial infections.
Louie, 2009
- Case series of first 1088 hospitalized and/or fatal cases of influenza A (H1N1) in California.
- April 23 – August 11, 2009.
- 32% were children < 18 years. Median age of all cases – 27 years.
- Overall rate of hospitalization or fatality per 100000 = 2.8 / 100000
o 11.9 / 100000 in infants < 1year.
o 1.5 in patients aged 70 years or older.
o Overall fatality | hospitalization = 118 / 1088 (11%), after median 12 days from symptom onset.
o See Fig. 2. Hospitalization rates highest in = 50 years, consistent with seasonal influenza.
- Chronic diseases in 741/1088 (68%).
- In 258 adults (20+) with known BMI, 156 (58%) were obese. 67 (26%) were morbidly obese. Some colinearity with other underlying conditions, since 103 of 156 obese cases also had chronic lung disease, cardiac disease, immunosuppression, and diabetes.
- Diabetes enumerated only for obese patients, detected in 31/103 obese patients. The minimum proportion of diabetes is therefore 31/258 = 12%.
- ICU | hospitalization = 340/1088 (31%).
- Ventilation | ICU = 193/297 (65%) of ICU cases with available information.
- Evidence of secondary bacterial infection = 46/1088 (4%).
- Point of reference – the percentage of adults who are morbidly obese in the US is 4.8% (Ogden, Carroll, Curtin, McDowell, Tabk, and Flegal, JAMA 2006).
- More than one third of adult cases reported nausea or vomiting, and one-fifth reported diarrhea.
- Other differences from seasonal influenza
o Age distribution of composite hospitalization or fatality uniform except for infants in this series – median age = 27 years.
o Those aged 60 years or older may have preexisting immunity.
- RF for disease
o Traditional – pregnancy
o Emergent – not previously independently associated with seasonal influenza – obesity and hypertension – however, may be proxy for other underlying conditions.
- In contrast with the perception that pandemic 2009 influenza A (H1N1) infection causes only mild disease, hospitalization and death occurred at all ages, and up to 30% of hospitalized cases were severely ill.
DL comment
- Louie et al., 2009 provide some evidence to support the conclusion that the virulence of swine flu virus is mild, and the mortality rates very low, compared with seasonal influenza.
From Molinari 2007
- Attack rates from literature: minimum and maximum among age/risk bands imputed.
o Working age adults: 6.6% (Range: 2.6% - 15.5%).
o Children under 5 years old: 20.3% (range 7.5%, 25.8%).
o Elderly individuals: 9.0% (Source? Unable to detect.)
- Death | flu infection
o Age 18-49 – P = 0.00009
o Age 50-64 – P = 0.00134
o Age >= 65 – P = 0.01170
- Hospitalization | flu infection
o Age 18-49 – P = 0.0042
o Age 50-64 – P = 0.0193
o Age >= 65 – P = 0.0421
- Death | hospitalization
o Age 18-49 – P = 0.021
o Age 50-64 – P = 0.069
o Age >= 65 – P = 0.278
Vaillant, 2009
- Analysis of available fatality data from numerous countries, until July 16, 2009, compiled by epidemic intelligence team at Institut de Veille Sanitaire. First death occurred in Oaxaca State, Mexico, with symptom onset on April 4, 2009.
- 126168 reported cases. Scarce data availability from African Countries. Data available for 574 deaths.
- Mean age was 37 years, most deaths (51%) occurring in 20-49 year-olds.
- Underlying disease, presence or absence, documented for 241/499 cases.
- Underlying disease | fatality = 218/241 (90%). This may be over-estimated since documented absence was required for a case to fit into the complement, as opposed to simple absence of documented disease.
- “Diabetes and obesity were the most frequently identified underlying conditions).
- Obesity in 7, diabetes in 5, obesity and diabetes in 1, and obesity and/or diabetes in 41 (some patients only had aggregate data available) for total 54 / 193 fatal cases (28%).
Kumar, 2009
- Case series of a multicenter cohort of critically ill adult and pediatric Canadian patients (report of the Canadian Critical Care Trials Group H1N1 Collaborative).
- Critically ill defined as admission to ICU or requiring mechanical ventilation, or FIO2 >= 60%, or need for IV inotropic or vasopressor medication.
- Prospective measurement with coordinating center ensuring data quality.
- April 16 to July 13, 2009.
- 215 critically ill patients.
- 168 with confirmed or probable A(H1N1) infection.
- Mean age in confirmed or probable 32.3 years. 25.6% of cases were aboriginal. A large cluster of cases were obtained from the greater Winnipeg region (52 patients).
- Comorbidities present in 165/168 (98.2%). Major comorbidities present in 51/168 (30.4%).
- Diabetes, type 1 or type 2, present in 35 /168 (20.8%).
- Ventilation | ICU = 136/168 (81%) during the first day of admission, most were ventilated invasively.
- Fatality | ICU = 29/168 (17.3%).
- As of August 22, 2009, 7107 cases of S-OIV H1N1 infection in Canada.
- Hospitalization | case = 1441/7107 (20.3%).
- ICU | case = 278 / 7107 (3.9%).
- ICU | hospitalization = 278/1441 (19.3%)
- Along the continuum of acuity of care, patients became older and were more likely to have one or more underlying medical conditions. Also more likely to be female.
- Severe disease and mortality concentrated among individuals aged 10 to 60, reminiscent of the W-shaped curve seen during the 1918 H1N1 Spanish pandemic. Few patients > 60 years old admitted to ICU.
- Diabetes and obesity, and other major underlying illnesses are collinear with First Nations status.
DL comment
- Kumar et al. calculate a much higher hospitalization | infection rate than expected for seasonal influenza, according to the estimates provided by Molinari et al., which include circulatory events likely to have been missed in this case series.
- This supports a higher severity of influenza for H1N1, but runs contrary to findings from Louie et al. which suggest that hospitalization rates are lower for H1N1 than those expected for seasonal influenza, as determined by Molinari et al.
- A possible source of bias is exposure suspicious bias, where those with severe influenza are more likely to receive diagnostic testing. The Kumar data likely over-estimate the hospitalization rate of infection, but to an unknown extent.
Jain, 2009
- Case series of patients hospitalized for H1N1 from May 1, 2009, to June 9, 2009 in the US. Retrospective chart abstraction, but all methods standardized.
- 13217 cases of infection and 1082 hospitalizations reported to US CDC.
- This case series describes first 272 completed chart abstractions.
- Median age = 21 years.
- Chronic medical condition in 198 of 272 patients (73%). 83% in adults.
- Asthma most common. Diabetes not enumerated.
- BMI available for 231 patients.
- Adult obesity in 29/100 (29%), morbid obesity in 26/100 adults (26%).
- ICU | hospitalization = 67/272 (25%).
- Fatality | hospitalization = 19/272 (7%).
- Underlying illness | ICU = 54/67 (67%).
- Ventilation | ICU = 42/67 (63%).
- Underlying illness | fatality = 13/19 (68%).
- Age distribution different from that of seasonal influenza
o Half of hospitalizations involved those under 18 years.
o Only 5% were aged >= 65 years.
o Possible explanations – increased case ascertainment in schools, etc.
- Other differences from seasonal influenza
o GI symptoms reported in 39% of patients.
o For seasonal influenza, 44 to 84% of adults hospitalized with influenza had an underlying condition. The H1N1 proportion fits into the upper range, at 83%.
- Obesity
o A majority of patients (81%) who were obese had an underlying condition associated with an increased risk of influenza-related complications.
o Prevalence of obesity similar among hospitalized adults and general population – but prevalence of morbid obesity much higher than the estimate 5% in the adult US population.
WHO. Comparing deaths from pandemic and seasonal influenza (). WHO, Geneva: 22 December 2009 (Accessed on October 11, 2010).
- Numbers of confirmed H1N1 deaths during the 2009 pandemic are sometimes compared to numbers of estimated seasonal influenza deaths
- Such comparisons are not reliable.
- Seasonal influenza estimates use statistical models to calculate “excess mortality”
o From all causes
o Regardless of confirmed influenza infection
- Laboratory-confirmed deaths
o Doctors often do not suspect H1N1 infection and do not test.
o When testing confirms H1N1 in patients with underlying medical conditions, the deaths may be attributed to the medical condition.
o Some tests for H1N1 infection are not entirely reliable – false-negative results are a frequent problem.
o Comparable estimates of excess mortality are not available.
- H1N1 appears to affect a younger age group – comparisons of numbers of deaths mask the distribution of deaths among individuals.
Donaldson LJ, Rutter PD, Ellis BM, Greaves FE, Mytton OT, Pebody RG, Yardley IE. Mortality from pandemic A/H1N1 2009 influenza in England: public health surveillance study.
- Estimate mortality from pandemic A/H1N1 2009 influenza up to November 8, 2009.
- Deaths
o Prospective reporting by hospitals, with nil reports mandatory.
- Denominator
o Laboratory confirmed cases ascertained.
o Weekly estimation of incident case numbers using two primary care surveillance networks – 20 million patients in England registered with a GP – estimates for consultation rates for ILI – used to scale up laboratory confirmed cases with those in which laboratory confirmation were not sought.
o Assumed that 50% to 80% of symptomatic ILI did not consult with GP – used to scale up H1N1 cases for those not presenting to medical attention.
o Two week lag in death reporting allowed for.
o Estimates scaled up to UK national population by age and sex groups.
- Case fatality rate – deaths per case of H1N1 influenza – range reported in [], reflecting uncertainty in the consultation rate of symptomatic ILI.
o Overall – 26 [11, 66] per 100 000.
o Children aged < 1 – 30 [2, 260] per 100 000.
o Adults aged 45-64 – 65 [21, 200] per 100 000.
o Adults aged >= 65 – 980 [300, 3200] per 100 000.
- Population fatality rate – Highest in children aged 5-14, in adults aged >= 65 years, and in children aged < 1.
- Number of cases – Highest in those aged 5-14, 15-24, and 25-44.
- Proportion of cases with diabetes – 9%, none of whom had diabetes alone.
- Overall case fatality rate of 0.026% (range [0.011, 0.066]) is lower than most estimates – but is based on a different denominator, intended to provide a better estimate of the true incidence of symptomatic cases.
- Highest mortality rate in the elderly, following a J-shaped curve. No evidence of high case fatality additionally occurring among young healthy adults.
- Elderly – lowest incidence rate but highest case fatality rate.
- The first influenza pandemic of the 21st century is considerably less lethal than was feared.
Reed C, Angulo FJ, Swerdlow DL, Lipsitch M, Meltzer MI, Jernigan D, Finelli L. Estimates of the prevalence of pandemic (H1N1) 2009, United States, April-July 2009. Emerging Infect Dis, 15(12): 2009.
- In most jurisdictions, the large number of ILI samples from milder cases led to a redaction of the previous policy of increased laboratory diagnosis by May 12th – focus on hospitalized patients.
- It is clear that many cases of H1N1 were mild – the true number may never be known, due to limitations in surveillance methodology.
- Reed et al. report incidence over the first 4 months of the pandemic in the US, adjusted for multipliers representing (probabilistic credible range reported for overall figures) …
o Medical care seeking – 42-58% among those not hospitalized (assumed to be 100% among those hospitalized)
▪ 42% - 2007 BRFSS (Where you ill with the flu? Did you visit a doctor?)
▪ 52-55% - 2009 ILI survey in 10 states (See BRFSS questions)
▪ 49-58% - Delaware university survey (Online survey assessed health-seeking behaviors – combined with data from the campus health center – during a large H1N1 outbreak)
▪ 52% - Chicago community survey
o Specimen collection – 19-34% among those not hospitalized, 40-75% among those hospitalized.
▪ 25% - 2007 BRFSS
▪ 22-28% - 2009 ILI survey in 10 states
▪ 19-34% - Delaware university survey
o Specimen submission – 20-30% among those not hospitalized, 50-90% among those hospitalized.
▪ 26% - Delaware university survey
o Laboratory detection of H1N1 - sensitivity – 90-100%.
o Reporting of confirmed cases to CDC – inclusion in national statistics – 95-100%.
- Multiplier estimates were obtained from prior studies and recent investigations, including field data, unpublished community surveys on ILI and health-seeking behavior, and the BRFSS.
- Multipliers were executed using a probabilistic model.
- The median multiplier of reported to estimated cases was 79 – every reported case may represent 79 total cases, with 90% probability interval of [47, 148].
- The median multiplier of hospitalized cases was 2.7, 90% probability interval [1.9, 4.3].
- Most cases appeared in the 5-24 age group – this group also had the highest incidence rate per 100 000 (2196), followed by the 0-4 age group (1870). Elderly patients accounted for the smallest number of cases, both absolutely, and as a rate (107 / 100 000).
- Hospitalizations – most occurred in the 5-24 age group. The highest hospitalization rate was estimated for the 0-4 age group. Again, the lowest hospitalization numbers and rates were estimated for the elderly stratum, at 1.7/100 000.
- Estimated total number of symptomatic cases through July 2009 – 3.0 million, 90% probability range [1.8, 5.7]
- Estimated hospitalization rate – 0.45%, 90% probability range [0.16, 1.2].
- Ratio of deaths to hospitalizations during this period was 6% - CDC laboratory-confirmed surveillance data.
- Reported cases of laboratory confirmed pandemic (H1N1) 2009 influenza are likely a substantial underestimation of the total number of actual illnesses that occurred in the community during the spring of 2009.
- Assumptions – representativeness of parameter estimates, parameter estimates obtained from studies of ILI, missing those with milder symptomatic illness.
- The total number of pandemic H1N1 2009 cases in the US during April-July 2008 may have been up to 140 times greater than the reported number of laboratory confirmed cases.
CDC. Updated CDC estimates of 2009 H1N1 Influenza cases, hospitalizations and deaths in the Unites States, April 2009 – April 10, 2010 (). CDC, Atlanta, GA: May 14, 2010 (Accessed October 11, 2010).
- Reasons for undercounting influenza deaths every year
o States not required to report individual flu cases or deaths
o Infrequently listed on death certificates of those who die from flu-related complications.
o Many seasonal flu-related deaths occur after initial infection, due to secondary pneumonia or because of aggravation of an existing chronic illness.
o Lack of testing, or lack of timely testing.
o Diagnostic accuracy – some tests only moderately sensitive.
- On-going estimate of H1N1 cases, hospitalizations, and deaths using the corrections for under-ascertainment employed by Reed et al..
- Reed et al. did not model the reporting of deaths explicitly – rather, the same factors affecting hospitalization are assumed to affect death reporting, and deaths are estimated based on the proportion of deaths per hospitalized case observed in laboratory-confirmed surveillance data.
- From April 2009 to April 10, 2010 – the entire season – H1N1 – rounded estimates only
o Infections – Median estimate 61 million, range [43, 89]
o Hospitalizations – Median estimate 274 000, range [195000, 403000]
o Deaths – Median estimate 12470, range [8870, 18300]
- Adjusted for under-ascertainment, this suggests a crude hospitalized case fatality rate of 4.6%, and a crude overall case-fatality rate of 0.0214%.
Case series of influenza hospitalizations or deaths
Finland, 1942
- Case series of staphylococcal cases “of interest” appear during a period of influenza epidemic in Boston. Cases presented to Boston City Hospital, except for 2 cases of “especial interest” from elsewhere. All cases had adequate bacteriologic evidence of staphylococcal pneumonia.
- Smith and Poland report 3 of 66 cases with diabetes. However, I could only detect 2 diabetic patients in 42 cases of staphylococcal pneumonia. Among these, 18 patients with acute staphylococcal pneumonia with rapid and complete pneumonia did not have clinical characteristics tabled. 11 patients with severe staphylococcal pneumonia survived – one of these was diabetic. The other diabetic patient died in the group of 7 patients with acute pneumonia complicating influenza. 6 patients had fatal organizing and fibrosing pneumonia, leading to death 15 to 56 days after onset of severe pulmonary symptoms.
- Several miscellaneous cases of staphylococcal infections of the lung, as well as some focal infections, are reported additionally. Why these are not enumerated with the others is not apparent. This case series is highly suspect for selection bias.
Giles, 1957
- Necropsy results for a case series of 46 patients admitted for pneumonia, in which influenza was either the primary or a contributing cause of death according to clinical judgement (influenza was only isolated from 8/14 specimens tested). Patients were selected from 53 deaths at City General Hospital in Stoke-on-Trent during an epidemic with a high attack rate in September-October, 1957.
- Staphylococcus aureus was isolated from lungs or bronchi in 15 necropsies.
- 2/46 deaths occurred in patients with diabetes (4.3%).
- The fatality rate of patients admitted to hospital during the epidemic exceeded 25% despite intensive antibiotic therapy.
Martin, 1958
- Case series of deaths in Boston during the 1957 Asian influenza pandemic. 118 excess fatalities suspected, 32 cases reported out of 43 fatal cases suspected of being influenza-associated at the study sites. Cases had either a solid clinical link to influenza by case history, or laboratory evidence of influenza infection.
- 15 cases were influenza without bacterial complication, 11 were post-influenza staphylococcal pneumonia, and 6 cases were post influenza bacterial pneumonia, non-staphylococcal.
- 21/32 cases had chronic disease of “major proportions”. 4/32 cases were pregnant. 3/32 cases had a history of diabetes. All three diabetic patients were middle-aged males, aged 30-45.
Oseasohn, 1959
- Case series of deaths in Cuyahoga county during the 1957 Asian influenza pandemic. 33 deaths studied, selected systematically based on a clinical influenza-like syndrome preceding death. Virus detected in 25/33 cases.
- Case provides examples of encephalitis, fulminant infuenzal and staphylococcal pneumonitides, fulminant abacterial influenza pneumonitis, and influenza with pneumococcal pneumonia +/- bacteremia. Duration of illness was less than a week in most cases, and many cases were non-elderly.
- Staphylococcus pneumonia definitely an important bacteriologic agent.
- 25 patients died from respiratory disease. 6/25 had underlying chronic cardiopulmonary disease. 1/25 had diabetes.
- Case series data is useful, but may be selective for younger patients in whom pneumonia deaths appear atypical, or who may be more likely to present with classical influenza signs and syndromes – bias against detecting diabetes. Also, diabetes as an underlying disease may not have been detected without systematic efforts to make this diagnosis. Case ascertainment of diabetes may have been lower in the past generally, and type 2 diabetes may not be obvious from the medical history during a fulminant 2-day course of disease. Systematic data collection is necessary – evidence from recent H1N1 may be useful.
Stuart-Haris, 1950
- Case series of 85 influenza-linked deaths during an epidemic period, and 11 influenza-linked deaths during an inter-epidemic period. Cases and controls were similar by age and sex. Also examination of 22 fatal illnesses in Sheffield during the 1949 influenza epidemic.
- On death certification / medical records recounting the acute event, influenza correlated highly with other respiratory disease and with cardiovascular diseases. It appeared random as to whether influenza was recorded as contributing or primary cause of death, relative to the underlying condition. E.g.: Pneumonia mentioned in 40 instances, heart failure in 60 instances.
- Sheffield cases: 9/22 fatal illnesses obtained positive viral cultures. 1/9 viral deaths had diabetes recorded. 1/13 virus-negative deaths had diabetic coma recorded. OR = 1.50, p = 1.00.
Bisno, 1971
- Cohort of CAP patients during the 1968-1969 Influenza A(2) Hong Kong epidemic, admitted to hospital in Memphis. This included 98 medicine admissions.
- 83% of patients with confirmed LRTI during this period had serologic signs of influenza infection.
- Study mentions that ~three forths of patients with influenza-associated pneumonia had identifiable underlying medical disorders or were pregnant. DM was listed as the third most common category, 8 patients out of 88.
- A control year was selected – 1969-1970, no influenza epidemic. Only 29 medicine admissions over the same period. Underlying chronic illness, including diabetes mellitus, noted in 23 (80%) of patients. DM may be a RF for CAP admission in patients with or without influenza.
- Case fatality during the epidemic was 14/106 (13%).
Case series of diabetes deaths during influenza epidemics
Watkins, 1970
- Case series of 29 patients with ketoacidosis seen in eight weeks, concomitant in time with an influenza epidemic from 1969-1970. Influenza cannot be proved to be the precipitating cause, but circumstantial evidence was “considerable.” Earlier reference to FitzGerald et al., 1961, who suggested that 9% of all keto-acidosis cases were the result of respiratory infections – it is suggested by this study that influenza may disturb diabetic control more than others.
- 3/29 patients may have been type 2 (“non-ketotic diabetes, two being finally controlled with oral hypoglycaemics”) – none of these patients died.
- 7 patients died, although only 3 of these deaths occurred in the acute phase of acidosis (death due to hypokalemia) – the other causes of death were cardio/cerebrovascular, pulmonary emboli, and meningitis.
Studies of excess mortality in influenza
Collins, 1930
- Objectives:
o Study the course of recorded mortality from pneumonia and influenza.
o Identify excess mortality.
o Measure the excess.
o Study its distribution across geographic regions of the US.
o Study the movement of epidemics from one region to another.
- Weekly pneumonia and influenza death certifications from 95 US cities of >100 000 population taken to be geographically representative – generalize to cities only.
- Excess mortality estimated by removing expected seasonal mortality.
- Expected seasonal mortality estimated by median P&I death rates for the 7-year period 1921-1927 – smoothed by 5 period moving average models, including epidemic and non-epidemic periods (epidemic peaks included, but tamped down by non-epidemic values and by the moving average function).
- Assumed – based on the observed trends – that there is no secular trend in P&I death rates since 1920.
- Total excess epidemic deaths / 100 000 measured. Length of epidemic measured by the inverse cumulative distribution function [I(0.75), I(0.25)] for excess deaths, or, alternatively, the IQR for weeks of excess deaths. Influenza epidemic period defined as the period during which the rate of P&I deaths appeared “definitely above the median”, ending when the curve had returned to approximately the median rate.
- Time period – 1920-1929, in which 6 epidemics occurred. Excess mortality = 250 000 total for the entire US, extrapolating from the 95-cities data.
- Note the collocation in time of pneumonia and influenza peaks.
- Collins also known for pioneering the use of excess mortality to define the start and end-points of an influenza epidemic.
Collins, 1932
- Author noted that studies of influenza mortality examine deaths credited to influenza or pneumonia – clinical grounds for specificity assumed, screen out the effects of coincidental epidemics of other diseases. However, some evidence had emerged that excess deaths from all other causes were also increased during influenza epidemics.
- Objective
o Measure the excess for all other causes in addition to pneumonia and influenza.
o Provide evidence that the excess for all other causes is due to influenza infection, and not coincidental epidemics.
o Examine the contributions of other causes of death to excess epidemic mortality.
- Data – weekly death data from 35 cities from 1917-1929 – Weekly death data was fairly rare at the time, and was only available from 35 large cities – nearly 25 million inhabitants.
- Methods as in Collins, 1930. Seasonal trends for particular non-PI causes of death estimated by averaging values for the same period of time from the previous and following years (very back of hand calculations).
- Excess mortality due to non-PI causes substantial, shown to be quite contemporaneous with PI excesses. Time properties (peak day, 25th, 50th, 75th percentile of deaths, IQR) quite similar.
- Negative control – heat wave – week of June 10, 1925 – all-cause mortality excess with no counterpart in the P&I rates.
- Percentage of total excess mortality credited to causes other than PI – about 40% for all epidemics other than 1918-19 (8%) and 1920 (23%) (much lower percentage in those major epidemics).
- Death certificate analysis – work in 10 cities demonstrates that the ratio of underlying : primary-listed P&I deaths is 27% – while the ratio of other primary-listed : primary PI-listed deaths was 64% – Less than half of the excess deaths credited primarily to causes other than PI listed PI as a contributory cause.
- Similar analysis for particular non-PI causes of death. Peaks in total deaths for diabetes corresponded in time to influenza peaks, but were smaller and less definite than in the case of organic heart disease.
- For epidemics of 1922, 1923, 1926, spring 1928, and winter 1928-29, combined primary cause of death data is tabulated
o Organic heart disease – 46.4% of non-P&I causes – 48.4 / 100 000 excess mortality.
o Diabetes accounted for 6.3% of excess mortality from non-PI causes, and 6.5/100 000 excess mortality.
- While the numbers may seem few, they represent a large proportion of cause-specific expected death rates
o Organic heart disease – excess is 15.1% of the expected.
o Diabetes – excess is 18.8% of the expected.
- “The chief causes to which excess deaths from causes other than influenza and pneumonia are credited during influenza epidemics are organic heart disease, nephritis, cerebral hemorrhage, diabetes, respiratory tuberculosis, bronchitis, and puerperal conditions other than septicemia”.
- Coincidence of diabetes mortality and influenza suggests an association. Diabetes listed among the “chief causes” of excess deaths other than PI – may be the origins of the “high risk group” designation here. However, it cannot be concluded that diabetes is a high-risk group because there lacks a valid low-risk comparison. Additionally, diabetes deaths likely to be keto-acidosis deaths in type 1 DM. Absolute contribution to numbers of deaths low. See problems with death certificate reporting, especially of type 2 diabetes.
Stocks, 1935
- Objective: Identify the causes that show the “most pronounced sympathetic increase in mortality” during influenza epidemics and which are unaffected, and to seek a reason for this.
- Death certificate analysis, 1921-1933.
- Months of first quarters in each year (3 per year), divided into months with normal, mild-moderate, or severe levels of influenza epidemic (by number of influenza deaths – cut-points set at 2000, 4000, and 7000).
- Excess mortality expressed as a percentage increase over the expected disease deaths for that period. Expected disease deaths estimated by linear regression for a secular trend pinned by two points at either end of the period in question, each point being calculated as the average over 5 years (i.e.: point +/- 2.5 years) – Adjusted for secular trends – seasonal trends not necessary because data from the same month of the year is involved in each estimation.
- Table 2: Stratify on month – order months for each year by influenza severity – ordinal ranking of years for each month. Examine the trend in excess other disease deaths with increasing influenza epidemic severity – excess as a percent of expected.
- Fig. 2: Assess whether cold vs warmer weather is a confounder of the relation in table 2 by performing the analysis in Table 2 within strata of temperature.
- Table 3: Simple analysis, mean number of deaths from other diseases by increasing severity of influenza epidemic.
- Table 4: Same as in table 3, but analyzed by months bucketed into normal, mild/moderate, and severe epidemic categories. Influenza-attributable fraction of deaths calculated by subtracting mild/moderate and severe from normal numbers.
- Table 5: To assess adequacy of death certificate reporting.
- Table 6: Similar point as table 4, but subtracting peri-epidemic from epidemic quarters in each year, and then obtaining the difference for years with an epidemic vs years of low influenza. Extrapolation to 1933 to assess fit.
- During moderate epidemics, each 100 influenza deaths was accompanied by 190 other deaths in excess of expectation, consisting of 80 from respiratory diseases without mention of influenza (bronchitis and pneumonia), 13 from cerebral hemorrhage, 7 from whooping-cough, 6 from “old age” 5 from TB, 2 from mitral disease, 71 from other diseases of the heart and circulation, and 6 from other conditions.
- Possible explanations – lowered state of resistance to many disorders concomitant with influenza epidemics; consultation of a physician only when sequelae have developed, so that early recognition of ILI is not available; or influenza, though recognized, is not mentioned on death certification.
- Many disease show rises in deaths contemporaneous with the presence and increasing severity of influenza activity, even in analyses stratified for warm/cold, month of year, and with adjustment for secular trends.
- Diabetes was not one of them. For diabetes and certain other chronic diseases, death would be assigned to influenza, primarily, but assigned to the underlying disease if influenza had not been recognized. These influences have contrary effects on the behavior of death certificate frequencies, assuming additional or hastened deaths due to influenza, so these results cannot be interpreted.
Serfling, 1963
- Methods paper introduces methods for quantifying excess mortality and the epidemic threshold. Key developments:
o Use the method of partitioning mortality for prospective, in addition to retrospective, analysis.
o Use of number of deaths instead of death rate, to allow data from areas of population flux to be included.
- Age of low access to computing power.
- Mortality best indicator of influenza activity, despite 4-week lag behind increase in morbidity. Reporting of other quantities through surveillance channels, laboratory diagnostic tests, industrial and school absenteeism not ideal. Nowadays, these reasons may not hold.
- Tasks
o Determine and extrapolate secular trend
o Estimate seasonal variation
o Distinguish epidemiologically significant departures from expected weekly mortality from random variation from endemic levels.
- Excess is the additional amount after subtracting seasonal and secular trends.
- Previous methods for estimating expected mortality – yield seasonal curves reflecting distracting irregularities of the data.
o The point-to-point linear trend method, e.g.: in Stocks, 1955 – except with linear trend estimated for each calendar week over many years – 52 weekly lines with a common slope – typically group into quarters.
o Collins – moving average approach – adjust the median value.
- Suggested method – use a mathematical function to describe seasonal change.
o Secular trends may be estimated by simple linear regression.
o Seasonality requires Fourier terms. A single Fourier term is sufficient.
- Simple least squares regression on non-epidemic years to extrapolate expected mortality in epidemic years.
- Y = b1 + a1*t + a2*cos((2(pi)*t/(N) – b2)
- Suggested – a two stage regression, since doing it all in one go restricts the estimation to non-epidemic years only, whereas the secular trend may be estimated from non-epidemic months in epidemic years as well. Reduces extrapolation.
- Steps: Estimate secular trend. Remove secular trend from the data. Estimate seasonal change. Restore secular trend component.
- A method of performing this involving only simple addition and one division is provided – method of double integration.
- Need to distinguish epidemic increase from regular random variation. Standard deviation proposed – 2 weeks with 1.65xSD over the expected number of deaths.
- A high level for only a single week may be due to reporting lags, or transmission errors.
- Assuming null hypothesis – no epidemics – the chances of one or more instances in which the epidemic threshold is exceeded is 0.75 – assuming analyses restricted to 26 weeks representing winter and spring – 0.75 = 1-(1-0.05)^26 – single-tailed. The chance of two consecutive excesses over 26 weeks is approx. 0.10.
- For a moderate sized epidemic, the risk of failing to detect is 0.9 at the end of the second week, but only 0.1 at the end of the third.
- The Serfling method requires careful selection of non-epidemic months for secular trend estimation, non-epidemic years for seasonal trend estimation, and removal of outliers, e.g.: due to heat waves.
Housworth and Langmuir, 1974
- Analysis of excess mortality, building on the tradition of Farr, Frost, Pearl, Collins, and Serfling.
- 1957-1966 monthly deaths from US Vital Statistics.
- Expected mortality rates estimated as in Serfling, except with numerous Fourier terms and an additional quadratic term for the secular trend. Least-squares fit.
- Innovation – Relative intensity statistics. Additive standard deviations of the expected deaths overall and between mutually exclusive causes of death devised by regressing on the same non-epidemic time periods for all causes. Relative intensity, I, is the sum of standardized residuals over the length of the epidemic with a transformation of 1/root(m), where m is the number of continguous periods corresponding to an epidemic. Under the hypothesis of no epidemic, I is a normal variable with zero mean and variance equal to the number of contiguous time periods corresponding to an epidemic. Relative intensities allow comparisons of epidemic severity between epidemics, and severity of mortality between causes of death within epidemics. Comparisons can now be tested statistically. Major limitation: The expected mortality for the entire multi-epidemic period must be fit with a single model.
- Note that relative intensities may classify as significant a cause of death with fewer excess deaths than one classified as non-significant. Examine percent excess from expected to see why.
- Confirmed observation by Collins that respiratory deaths account for merely 40% of excess mortality. The proportion varied from 30 to 38% in more severe epidemics, to less than 25% in milder epidemics.
- The majority of excess deaths are due to heart, circulatory, and nervous causes, accounting for 44% to 62% of excess deaths during influenza A epidemics – exceeding 50% in all years except one.
- Heat wave – respiratory deaths significantly increased, but relative intensity not as high as those of other causes of death.
- Influenza B – Lower relative intensities for death for any cause, non-significant for “all other” category. Proportion of excess mortality due to heart, circulatory, and nervous causes much higher (61-70%).
- Relative intensities for diabetes were significant for all epidemics, and the heat wave, except the 1966 mixed A2/B epidemic. Relative intensities fluctuate – in 1960 diabetes relative intensity highest of all causes.
- Rates of excess mortality for diabetes range from 1.49-5.06 per million in the years in which relative intensity was significant. Rates of mortality were generally much lower than arteriosclerotic heart disease.
- Median proportion of all arteriosclerotic heart disease deaths related to influenza during influenza epidemics – 5.9%, range [4.1, 8.9].
- Future studies should examine mortality from all causes rather than respiratory disease deaths to capture excess mortality due to influenza.
- Severe influenza epidemics influence mortality from TB, asthma, chronic rheumatic heart disease, and diabetes to a small but statistically significant degree.
Alling, 1981
- Estimate excess mortality using new methods.
- Sabin had argued that the Serfling method deployed by the CDC overestimated excess mortality. While Sabin’s article is not entirely clear RE: mechanisms of bias, one suggestion is that the CDC estimates did not account for decreases in observed mortality following epidemics due to removal of sicker individuals earlier.
- New methods of estimation have been derived with advantages over Serfling.
- The differences:
o Instead of a Fourier term for seasonality, regression on indicators for month of the year.
o Addition of an indicator for influenza activity.
o The total number of deaths from which expected deaths are subtracted were actual observed values. Close correlation of actual observed values with values for total mortality predicted from the regression equation was used as a diagnostic for the regression.
- These two differences allow the expected mortality to be estimated using all data, as opposed to only non-epidemic data. The indicator for influenza activity could be set to zero (absence of influenza) or to a value equal to the influenza activity in a non-epidemic year. The chosen indicators here were monthly number of deaths due to acute respiratory causes, or influenza ICD-coded deaths.
- Mortality data – 1968-1974 – National Center for Health Statistics – Vital Statistics
o Respiratory = acute bronchitis and bronchiolitis, influenza, and pneumonia.
o CVD = active rheumatic fever, chronic rheumatic heart disease, HTN disease, ischemic heart disease, and other forms of heart disease.
- Disadvantage – does not account for qualitative effects, such as antigenic drift.
- Average excess deaths: 13000 (all ages), 9500 (elderly), 4500 (CVD), 6800 (Acute respiratory).
- Percent excess of expected cause-specific mortality was 10x higher for acute respiratory than cardiovascular, and relatively low for all causes.
- Estimates of the proportion of total excess mortality due to different causes seemed to vary much more from year to year than in previous data.
- Data show that the excess mortality from all causes of cardiac disease is a significant component of total excess mortality occurring during influenza epidemics – reasons obscure, phenomenon appears to be real – subject deserves more investigation.
- The excess mortality in the elderly is estimated at 1 in 2200 persons in an average influenza year – small risk – requires high powered studies to examine – implications for vaccination – further evidence is required before it can be concluded that the excess mortality in the elderly associated with influenza can be reduced by the routine use of vaccine.
Carrat, 1995
- Objectives – estimate influenza-associated mortality for a variety of death certificate causes of death, and estimate rate of deaths avoided by vaccination.
- Death certificate data, analysis proceeded by modified ARIMA to account for serial auto-correlation. Linear component of deaths regressed against influenza activity, with a different regression equation estimated for each year. Unsure how seasonality is corrected for in this model. Deaths from 1980-1990. Associated component was small or non-significant for non-epidemic years – good negative control. Test of fit by average residuals good.
- Rate of deaths avoided estimated using a formula and a sensitivity table.
- The number of deaths attributable to influenza ranged from 0.5 to 8.2 times the number of deaths registered as due to influenza alone. Attributable deaths especially important for respiratory diseases, cardiovascular disease, and chronic renal failure. Less so for diabetes and lung cancer.
- Diabetes: Significantly associated influenza component in only 2 of 5 epidemic years, associated rate = 6.2/100 000 and 5.6/100 000 compared with 14.7 to 29.8 for ischemic heart disease and 16.1 to 40.7 for pneumonia.
- However, death due to diabetes may have been restricted to keto-acidosis in type 1 patients, which is less common in any event. Authors include diabetes in the general listing of “high risk conditions” with increased risk of death from influenza. NOTE however that denominator = general population, and that there is no valid comparison to demonstrate an increased risk of death in high risk patients – compared to who?
Thompson, 2003
- Produce age-specific estimates of deaths attributable to influenza, by virus type and subtype, and to RSV.
- Three death categories modeled – underlying PI deaths, underlying respiratory and circulatory deaths, and all-cause deaths.
o PI most specific, most highly correlated with influenza activity.
o All cause traditionally used to estimate total burden.
o Underlying respiratory and circulatory deaths suggested to be a useful compromise between sensitivity and specificity – excludes deaths due to fires and motor vehicle crashes.
- Poisson regression (log link with offset determined by census data).
o Secular trend captured by including a linear and quadratic time term.
o Seasonality captured with sinusoidal terms.
o Indicator of viral activity = proportion of specimens testing positive.
- Data sources
o 1976-77 to 1998-99 seasons, mortality data from Vital Statistics (NCHS).
o Population data from US Census.
- Testing results
o Average 27360 specimens tested per year, average of 12% of specimens positive for influenza.
o Average of 16% of 107711 specimens tested positive for RSV.
- Numbers of deaths in each category increased over time
o e.g.: PI deaths increased by 83% - more than the trend for circulatory or all-cause deaths.
o This is likely due to the aging of the US population, with a doubling of persons aged 85 years or older – these patients had RR 14.8 [14.6, 14.9] more likely to die of an influenza-associated all-cause death than persons aged 65 to 69 years.
o Mortality has not increased over time for patients in 5 year age bands > 65 years old (e.g.: 65 to 69, 70 to 74, etc).
- Average yearly deaths
o PI – 8097 (SD 3084) – 9.8% of all P&I deaths
o Respiratory and circulatory deaths – 36155 (SD 11055) – 3.1% of all such deaths.
o All cause deaths – 51203 (SD 15081) – 2.2% of these deaths.
o Significant increases over the study period for each category.
o Influenza-associated deaths represented 9.8% of total PI deaths, 3.1% of respiratory and circulatory deaths, and 2.2% of all-cause deaths.
o RSV was associated with a substantial, but smaller, proportion of these deaths.
- Rates, average
o PI – 3.1 / 100 000 person years
o Circulatory and respiratory – 13.8 / 100 000 person-years
o All-cause – 19.6 / 100 000 person-years
o Influenza death rates were ~3 times those of RSV (p < 0.05 for each RR), although there were age groups in which the RSV rates of death were much higher (age = 90 years.
- Majority of patients (80%) whose death was attributed to influenza had chronic heart or chronic lung conditions, with chronic lung disease present in >50% of all patients.
- 350 deaths attributed to influenza among persons with diabetes – 86% had co-existing chronic lung or heart disease.
- For persons aged 65 years and over, the RR of influenza-attributed death = ~20 for those with both chronic heart + chronic lung disease than for those without either conditions. RR ~12 for patients with chronic lung disease, RR ~ 5 for patients with chronic heart disease.
- Table 1 has complete age-specific rates of death and hospitalization for different co-morbidity groups (chronic lung and heart disease only).
- Hospitalized case fatality rate due to influenza increased from 4% (age 50-64) to 30% (aged 90 years or older) – approximately similar to the hospitalized case fatality rates calculated from Molinari et al..
- Age as well as co-morbidity contribute independently to elevated risk of influenza-associated mortality.
- This study lacked estimates of precision due to lack of precise prevalence estimates for risk conditions. Unknown accuracy of comorbidity coding due to use of CIHI DAD – were they only able to use comorbidity data from the admission leading to death? Collection of hospital deaths may also miss deaths in the community that did not result in hospital admission – these deaths may be incomplete for comorbidity data.
- Note the small number of estimated influenza-associated deaths in diabetes.
Studies of outcomes in CAP
Fine, 1996
- Objective – review the medical literature on the prognosis and outcomes of patients with CAP.
- Systematic review and meta-analysis, covering studies from Medline up to June 1995.
o Studies were included if they concerned CAP with radiographic confirmation in at least 50% of cases.
o Primary outcome was mortality (at any given follow-up time). The review also attempted to track a variety of non-fatal morbidities, such as symptom resolution, empyema, and satisfaction.
o McMaster scale (scored out of ten) used to display quality, does not seem to have informed analysis. Sensitivity analyses with outliers dropped affected some results.
- Analysis
o Calculation of univariate summary ORs of death for relevant prognostic factors.
o ORs pooled using the method of Mantel and Haenszel.
- 122 studies included, reporting 127 cohorts of CAP patients with 33148 patients.
- Overall mortality 13.7%.
o 5.1% (6 cohorts) in hospitalized and ambulatory patients
o 13.6% (84 cohorts) in hospitalized patients
o 17.6% (9 cohorts) in elderly
o 19.6% (12 cohorts) in bacteremic patients
o 30.8% (6 cohorts) in nursing home patients
o 36.5% (13 cohorts) in ICU patients.
- Age significantly associated with mortality – OR = 1.05 [1.01, 1.09] for each 10 year increment by logistic regression of mortality on mean age reported in each study.
- Significantly associated with mortality by summary ORs: male, chills (protective), pleuritic chest pain (protective), altered mental status, dyspnea, tachypnea, hypotension, hypothermia, CHF, alcohol abuse, diabetes mellitus, immunosuppression, cancer, coronary artery disease, neurologic disease, leucopenia, bacteremia, multilobar infiltrate, and azotemia.
- Diabetes: 5 cohorts, 14655 patients – OR = 1.3 [1.1, 1.5] – RD = 2 [1, 4].
- Oddly, tobacoo use and prior pneumonia were non-significant.
- Mortality also highly associated with bacterial etiology. Study also demonstrated 9% mortality in influenza A pneumonia (combining all patient Ns).
- Studies selected for more severe pneumonia cases. Most concerned hospitalized patients although most CAP is treated in the ambulatory setting with short-term mortality of less than 1%-2%.
- Only 22% of studies reported follow-up time for the mortality outcome – biases possible.
Houston, 1997
- Objective – Assess factors associated with 30-day mortality in a population-based study of older adults with LRTI.
- Retrospective cohort study
o Cohort of pneumonia patients identified from databases of the Rochester Epidemiology Project, which contain standard health information from all major health care providers (a small number, actually) in Olmsted County.
o First LRTI identified by diagnosis without requiring radiography or positive laboratory results “because this was a retrospective study of actual practice”.
o LRTI included confirmed pneumonia, suspected pneumonia, and bronchitis.
o Patients followed for one year (1987) for case identification.
- Outcome = 30-day mortality, to increase sensitivity and minimize potential differences in length of stay.
- Covariates determined from medical record review, including demographics, previous medical history, smoking, and medications. ONLY the FIRST LRTI eligible.
- Interesting that a minority of deaths (16%) actually have pneumonia written on death certification as the underlying cause of death. Is this a methodologic failing? 30-day mortality was chosen for “sensitivity” to pneumonia-induced mortality, but does this reflect adequate specificity? Discussion suggests the possibility of an LRTI induced “cascade”.
- Population included community and nursing home residents.
- Univariate significant: Older age, atypical symptoms (poor eating, confusion, or lethargy), cardiac disease, neurologic disease, current cancer, hip fracture in previous 2 years, renal failure, recent or current use of antibiotics, tranquilizer use, steroid use, immunosuppressive drug use, and nursing home residency.
- Multivariable significant: atypical symptoms, neurologic disease, current cancer, use of antibiotics (28% for index LRTI, 28% for previous LRTI or COPD, 29% for UTI prophylaxis, 15% for other indications).
- Diabetes – prevalence = 6%, univariate OR = 1.66 [0.54, 5.07].
- Age NS after adjustment for other covariates. Underlying pulmonary illness NS – possibly due to a survivor bias in cohort selection. Or possibly more LRTI incidents, but no increased risk of death for any one incident.
Akbar, 2001
- Determine the causative organisms, antimicrobial susceptibility, and mortality of community- and hospital-acquired pneumonia in diabetics.
- Cohort study of CAP or HAP cases identified by sputum culture test positive for bacteria, 1998-1999, at King Abdulaziz Hospital.
- 354 cases included, 125 of whom had diabetes. Diabetes patients were older, more likely to be male. 85/354 cases were CAP, 269/354 cases were hospital-acquired.
- No significant differences in mortality. Overall mortality 45%. CAP mortality – diabetic patients = 8/26 (31%), non-diabetic patients 12/59 (20%).
- Very atypical spectrum of microorganisms dominant, with H. influenza first and Staphylococcus aureus / Moraxella catarrhalis second. Diabetic patients more likely to be infected with Staphylococcus aureus than non-diabetic patients (23% vs 10%, p = 0.02).
- Case selection biases
o Cases started on antibiotics without definitive evidence do not show up. May select against regular streptococcus pneumonia pneumonias.
o More invasive procedures more likely to test positive – however these may be reserved for more complicated pneumonias.
o Many of these cases were HAP, and the most common HAP pathogens are Staphylococcus and gram negative bacilli. However, study does not stratify findings for these pneumonias.
- Low quality analysis, without adequate control of confounding or adequate definition of the cases.
- Lipsky, Diabetes Care, 1987 demonstrated higher rates of nasal carriage of Staphylococcus aureus in diabetic patients, providing a possible mechanism for the spectrum of microorganisms.
Falguera M, Pifarre R, Martin A, Sheikh A, Moreno A. Etiology an outcome of community-acquired pneumonia in patients with diabetes mellitus. Chest, 128(5): 2005
- Check references 9, 10.
- Objective – Determine whether the clinical or radiologic findings, the causative microorganisms, or the outcome of CAP are modified by the presence of diabetes mellitus.
- Cohort study
o Prospective diabetes registry, cases presenting the emergency room at the Arnau de Vilanova Hospital in Lleida (Catalonia, Spain), 1998-2002.
o Radiographically confirmed CAP.
o Diabetes detected by previous diagnosis, treatment with oral drugs, or by hyperglycemia detected during the current episode.
- Outcomes and analysis
o Mortality and pleural effusion
o Univariate analyses stratified on PSI.
o Multivariate analyses do not include PSI because of overlap with other variables.
o All candidate variables were selected for multivariable, although it appears that only positive results are reported.
- 660 patients included. 324 with underlying diseases, 106 with diabetes, 95% of which was type 2. Diabetes patients older and had more severe pneumonia, with more frequent hospitalization. Also more frequent pleural effusions. Not significantly more empyema or complicated parapneumonic effusion.
- Diabetes and mortality
o Univariate – 17% vs 8%, p = 0.002.
o Multivariate – OR = 2.137 [1.090, 4.189].
o Other significant RFs – multilobar infiltrate, concomitant underlying diseases, bacteremia, elderly, and empyema or complicated effusion associated with increased mortality.
- Diabetes and pleural effusion
o Multivariate – OR = 2.005 [1.227, 3.277].
- RFs for mortality among patients with diabetes
o Univariate – Mortality associated with concomitant underlying diseases, diabetes complications, and multilobar infiltrate.
o No relation with duration of diabetes, glucose level, HbA1c, insulin therapy during pneumonia, bacteremia, PE, or empyema.
o Multivariate – only multilobar infiltrate significantly associated with mortality.
o Suggestive that the diabetes effect may be mediated by complications, or unadjusted confounding (-dl comment).
- Microbiological results do not suggest a different spectrum of infectious agents.
McAlister FA, Majumdar SR, Blitz S, Rowe BH, Romney J, Marrie TJ. The relation between hyperglycemia and outcomes in 2,471 patients admitted to the hospital with community-acquired pneumonia. Diabetes Care, 28(4): 2005.
- Objective – Examine the relationship between hyperglycemia and short-term outcomes.
- Prospective cohort
o CAP database at Capital Health, in Edmonton.
o November 2000 to November 2002 (2 years) – adults admitted with a clinical diagnosis of CAP in six Capital Health hospitals.
o All patients treated in a standardized manner according to a regional critical care pathway for CAP inpatients. Baseline data collection in the ED and by an in-patient physician. Standardized data collection and pathway implementation facilitated by rained research nurses.
o Analysis of 2471 patients with complete glucose measurements.
- Outcomes – In-hospital – death, non-metabolic complications, cardiac complications, nosocomial infections.
- 2471 patients, most of whom had admission glucose 14mmol/L (t2dm patients) and in those >6mmol/L (non-diabetic patients).
- Differences in the North Jutland vs Aarhus populations may have manifested in contrary results – North Jutland sub-cohort was older and had more comorbidities.
- Patients with type 2 diabetes have an increased risk of death associated with pneumonia hospitalization. Glucose on admission is a very important clinical indicator among patients with pneumonia, and may mediate a substantial part of diabetes-associated mortality. The diabetes-related effect may also be mediated by underlying renal disease.
Gamble JM, Eurich DT, Marrie TJ, Majumdar SR. Admission 2010
- Objective – Estimate rates and potential associations between hypoglycemia at the time of admission for CAP and all-cause mortality in-hospital, at 30 days, and at 1 year after admission.
- Cohort analysis of patients admitted to hospital with CAP. Subjects with casual blood glucose greater than 6.1mM excluded. Casual blood glucose categorized into 20 cigarettes / day, previous respiratory infection, chronic bronchitis
o Benzodiazepines - protective.
- Advantages – Population based, avoids potential selection biases. Small area, likely that all cases were identified.
- Limitations – uncertain if the risk set is updated regularly.
- Study provides new and more established evidence on the factors associated with the occurrence of pneumonia in the adult community.
Benfield, 2007
- Aim – Test the hypothesis that diabetes and hyperglycemia influence susceptibility to, and the outcome of, infectious disease hospitalization.
- Prospective cohort study
o Cohort of Copenhagen Heart Study patients – adults >= 20 years old.
o Danish national databases used for mortality and morbidity measurement.
o Exposure history assessed by baseline survey.
- Exposure – diabetes or hyperglycemia (but not in the same model).
- Outcome = first hospitalization with an infectious disease, censored thereafter. A second analysis involving hospitalized patients took death within 28 days as the outcome. Cases validated – positive culture and clinical signs required.
- Analysis – CPH models, univariate first, with variables significant at p < 0.10 forwarded into multivariable modeling.
- 10063 patients, 353 of whom had diabetes. 71509 person years of follow-up over 9 years (maximum).
- 90 infectious disease hospitalizations in patients with diabetes, and 1104 infectious disease hospitalizations among patients without diabetes.
- Most common infections were pneumonia*, UTI*, skin*, diarrhea, sepsis*, other viral, mycoses, URI, TB, Hep, HIV/AIDS, meningitis, and parasitism. (* Appeared more frequent in patients with diabetes, therefore forwarded into hazard ratio analysis.)
- Pneumonia hospitalization
o Diabetes HR = 1.75 [1.23, 2.48], p = 0.001
o Adjustment for age, sex, cholesterol, HTN, income, education, smoking, physical activity, and lung function.
o No vaccination status available.
- UTI – diabetes HR = 3.03 [2.04, 4.49] after adjustment for similar covariates (without lung function, including alcohol use).
- Skin infectious – diabetes HR = 3.43 [1.50, 3.95] after adjustment for similar covariates (without lung function, including alcohol use).
- Sepsis
o Diabetes HR = 3.40 [1.80, 5.30]
o Adjustment for similar covariates, including lung function and alcoholism.
o Considered non-significant after Bonferonni’s correction.
- Only hospitalization due to UTI was associated with increased risk of death in 28 days.
o Diabetes – 12.1% vs 3.4% – p = 0.037
o HR = 3.90 [1.20, 12.66]
o No covariates in model due to no covariates significant at p < 0.10 univariate.
- Plasma glucose (continuous) associated with increasing rate of hospitalizations from any infection, as well as increased hospitalizations from pneumonia, UTI, and skin infections.
o Magnitude of the HRs for these associations smaller than the diabetes HR.
o Each 1mM increment associated with a 6-10% increased adjusted RR for each of pneumonia, urinary tract infection, skin infection, and sepsis (p = 0.09, all other p-values significant).
o Diabetes was not entered into these models.
- “Diabetes and hyperglycemia are strong and independent risk factors for pneumonia, UTI, and skin infection. Furthermore, diabetes has a negative impact on the prognosis of UTI.”
- Review references for this paper in introduction and mechanistic references in discussion.
Houston, Journal of the American Board of Family Practice, 1995
- Documented that 52% of LRTIs in community-dwelling elderly adults was treated in the community setting.
- Use this to demonstrate one of the deficiencies of the current pneumonia outcomes literature.
Diabetes and other infectious diseases
Thomsen, 2004 (Diabetes Care)
- Cohort of patients with invasive pneumococcal disease in a registry in Jutland County, Denmark. Diabetes determined from drug prescriptions and previous hospitalizations, as well as from plasma glucose levels taken during the episode of invasive pneumococcal disease (sensitivity analysis with these patients removed to account for survival bias).
- 628 patients from 1992-2001.
- Diabetic patients were older and had more CVD.
- Mortality at 30 days was 11.1% vs 16.5% (diabetes vs no diabetes). Crude HR = 0.7 [0.3, 1.4]. At 90 days mortality = 16.0% and 19.5%. Crude HR = 0.8 [0.4, 1.5]. Adjusted HR not much different = 0.6 [0.3, 1.2].
- Possible that diabetes patients had cases detected in less severe / earlier form? Comparison of severity suggests higher CRP, higher severe sepsis, and comparable bacteremic loads, as well as comparable provision of a correct diagnosis and institution of appropriate antibiotic therapy in diabetic patients vs no-diabetic patients.
- Study provides evidence that diabetes is not associated with worsened outcomes of invasive pneumococcal disease.
- No vaccination data (but authors report low rates of vaccination in general). Charlson used to adjust statistics.
- Some interesting mechanistic references.
Background information for influenza
Sullivan KM, Monto AS, Longini IM. Estimates of the US health impact of influenza. American Journal of Public Health, 83(12): 1993
- Objective – Estimate the burden of influenza based on weekly identification of the onset of respiratory illness, and determination of infection by virus isolation and serology.
- Cohort study of 10% of the population of Tecumseh.
o Weekly telephone contact to check for respiratory illness. All such illnesses recorded, serologic and naso-pharyngeal specimens for viral isolation taken.
o All patients subject to routine blood samples every 6 months.
o 1977-1981 – 5 influenza seasons.
- Analysis - Rates standardized to 1980 US population.
- Clinical attack rate ranged from 5.4% in 1981 (A(H1N1)) to 20.4% in 1978 (A(H3N2)). Attack rate highest in those aged < 20 years old.
- Excess respiratory illnesses/1000 due to influenza estimated at average 13.8 to 16.0 per thousand.
- For the younger age group, the total number of respiratory infections during influenza season was 6.4 times and 5.6 times higher than the lower and upper estimates of excess influenza (based on sens and spec of diagnostic tests).
- For Those aged 20+, the number of total respiratory infections was 26.4 and 24.6 times higher than the lower and upper estimates of excess influenza.
- Here “excess respiratory illness” does not refer to the concept of excess mortality, but, rather, to any respiratory illness with laboratory confirmed influenza, not just ILI-syndrome. The excess is the number of respiratory events in those with influenza minus the number in those without.
Govaert, 1994
- Estimate the efficacy of influenza vaccination in elderly individuals using clinical and serological outcome parameters.
- Single influenza season, 1991-1992. Patients selected from lists of 34 family physicians in 15 practices in the Netherlands. Patients aged 60 years or more. 1838/9907 eligible agreed to participate (927 vaccine group, 911 placebo group; at analysis, 902 vaccine, 889 placebo group (drop outs removed for serology measurement only)). Patient-blinded placebo-randomized controlled trial. Randomization stratified by disease categories (cardiac disease, pulmonary disease, diabetes mellitus, and other/healthy). Blood samples taken at 3 weeks (S1) and 6 months (S2). S2 vs S1 4-fold increase in titre indicative of infection. Physicians asked to register ILI and provide blood samples. Patient-reported ILI also assessed by questionnaire after 10 weeks and 23 weeks. Questionnaire follow-up was 96%.
- HIGH-RISK GROUPS were excluded (e.g.: heart or lung conditions, diabetes, chronic renal insufficiency, and chronic staphylococcal infections.) – (THEN WHY WERE THERE STILL PATIENTS IN “HIGH-RISK” STRATA SINGLED OUT FOR ANALYSIS??)
- Placebo group outcome rates
o Serology: 80/911 (9%)
o Family physician: 31/911 (3%)
o Sentinal stations (?? – Is this a reference to the criteria used to judge patient questionnaires for cases?): 89/911 (10%)
o International classification of health problems in primary care-2 defined criteria, judged by family physicians: 129 (14%) (WHY IS THIS VALUE LARGER THAN “FAMILY PHYSICIAN”?) (ICHPPR-2 “considered the most lax”)
o Combined clinical and serological: 38/911 (4.2%)
- RR
o Serology: 0.50 [0.35, 0.61]
o Family physician: 0.53 [0.39, 0.73]
o Sentinal Stations: 0.69 [0.50, 0.87]
o ICHPPC-2 defined: 0.83 [0.65, 1.05]
o Combined clinical and serological: 0.42 [0.23, 0.74]
o Effectiveness similar for all sub-groups. However, sub-group confidence intervals quite wide (NS in most cases). As well, diabetes not reported separately from chronic heart and lung disease.
- The results of this study are consistent with a halving of the influenza risk by vaccination. Note: Removal of false positives by limiting period of analysis to the epidemic period increased RRs and precision for some outcomes.
- Age >70 may be an effect modifier (potential decrease in effectiveness).
Edwards, 1994
- Randomized controlled trial of cold-adapted and inactivated vaccines among non-elderly adults (triple-arm study, double-blinded).
- Efficacy outcomes: ILI, cultured ILI, culture-positive ILI.
- Study cohorts followed over 5 years.
- Control group:
o ILI, retrospectively reported at Spring: 240/1125 (21%), 262/1064 (25%), 146/1016 (14%).
o ILI, presenting for culture: 92/878 (10.5%), 119/1125 (10.6%), 125/1064 (11.7%), 93/1016 (9.2%).
o ILI, positive cultures: 28/878 (3.2%), 32/1125 (2.8%), 29/1064 (2.7%), 18/1016 (1.8%).
Keitel, 1997
- 30-60 year old healthy volunteers.
- Randomized to trivalent inactivated influenza virus vaccine or placebo. Placebo group during initial year of participation re-randomized until they received vaccine. Multivac group progressively increased in size, recruitment performed every year to maintain placebo pool. Patient blinded trial. Followed for 5 influenza seasons.
- Outcomes: Prospective and retrospective cases of influenza, both laboratory confirmed (serology for retrospective cases). Outcomes stratified by influenza season and virus subtype.
- Placebo virus isolate
o 1983-84: 3.4% H1N1, 2.7% B
o 1984-85: 2.9% H3N2
o 1985-86: 6.3% B
o 1986-87: 4.6% H1N1
o 1987-88: 4.8% H3N2, 2.1% B
Bridges, 2000
- Randomized controlled trial of influenza vaccination among non-elderly adults over two influenza seasons.
- Placebo laboratory confirmed illness rate was 6/137 (4.4%) in 1997-98, and 14/137 in 1998-99.
Allsup, 2004
- Determine the cost-effectiveness of influenza vaccination for healthy community dwelling individuals aged 65 to 74 years old in the UK.
- Recruitment from GP lists through September-October 1999, with treatment administered October-November. Patient-blinded placebo controlled RCT. Influenza trivalent vaccine. Follow-up over 1 influenza season only (1999-2000). Primary outcome was GP diagnosed ILI or pneumonia. Secondary outcome was patient self-reported ILI.
- A(H3N2) epidemic with peak rate of 278/100000. Good vaccine match.
- 729 patients randomized, 552 to vaccine, 177 to placebo. No LTFU over the year.
- Placebo arm GP diagnosis of ILI occurred in 1.1%. GP diagnosis of pneumonia occurred in 0%.
- Placebo arm “at least one or more episodes of self-reported influenza like illness” occurred in 8.9%.
- No laboratory confirmed ILI sought. GP diagnoses may not be sufficiently sensitive.
Molinari, 2007
- Economic analysis of annual impact of seasonal influenza in the US.
- Payer and social perspectives.
- Costs estimated for five age groups, and for high-risk and low-risk medical groups.
- Costs estimated for 4 scenarios: ill but not medically attended, ill with outpatient care, ill with hospitalization, ill followed by death.
- Incident cases
o Age-specific excess hospitalizations and deaths estimated using regression methods from studies like Thompson et al.. Deaths and hospitalizations from PI and from respiratory and circulatory conditions only considered, respiratory and circulatory events considered the primary outcome for analysis. Identical risks assumed for low and high-risk conditions. Data covered 1980 to 2001.
o Outpatient visits due to influenza estimated from Hurwitz, Haber, Chang, et al., JAMA 2000; Menee, Black, MacWilliams, and Aoki, Can J Public Health 2003; and other studies, most of which are trials. These two references are the adult primary references.
o Prosser, Bridges, Uyeki, et al., Emerg Infect Dis 2006 suggests that high-risk patients attend for outpatient ILI at a rate twice that of low-risk patients.
o Non-attended cases estimated from the complement, i.e.: total number of cases minus hospitalized, fatal, and outpatient attended cases. All numbers / year determined by multiplying P across the US population.
o Attack rates from literature: minimum and maximum among age/risk bands imputed.
▪ Working age adults: 6.6% (Range: 2.6% - 15.5%).
▪ Children under 5 years old: 20.3% (range 7.5%, 25.8%).
▪ Elderly individuals: 9.0% (Source? Unable to detect.)
o Death | flu infection
▪ Age 18-49 – P = 0.00009
▪ Age 50-64 – P = 0.00134
▪ Age >= 65 – P = 0.01170
o Hospitalization | flu infection
▪ Age 18-49 – P = 0.0042
▪ Age 50-64 – P = 0.0193
▪ Age >= 65 – P = 0.0421
o Death | hospitalization
▪ Age 18-49 – P = 0.021
▪ Age 50-64 – P = 0.069
▪ Age >= 65 – P = 0.278
o Gross attach rate in kids over twice as high as that of elderly adults – hospitalization rate lower than elderly adults, comparable to that of non-elderly adults, and much higher than that of adolescents / young adults – death rates very low.
- Hospitalized days and total days of productivity lost similarly calculated (sources not reported).
- Costs
o Estimated using the Medstat Marketscan (database of privately insured individuals, including Medicare-eligible individuals with supplemental insurance) databases for cases of influenza in the three medically attended categories (ICD-9 and ICD-10 codes PI codes for greater specificity). Costs inflated to 2003 prices using the medical care component of the CPI.
o Cases not medically attended: Assumed OTC medication direct costs $3 / case.
o Indirect costs = lost productivity, each outpatient visit per patient assumed to represent a day of productivity lost per case. Hospital days = day of productivity lost. Non-medically attended cases assumed to be 0.5 LP for younger adults, 1 LP for elderly.
o Costs per case (age 50-64 and 65+), LP = lost productivity, dollar amount refers to direct medical costs.
▪ Not medically attended: $3, 0.5 LP; $3, 1.0 LP.
▪ Outpatient visit in non-high risk patients: $150, 2 LP; $242, 3 LP.
▪ Outpatient visit in high risk patients: $733, 4 LP; $476, 7 LP.
▪ Hospitalizations in non-high risk patients: $22304, 13 LP; $11451, 13 LP.
▪ Hospitalizations in high-risk patients: $31309, 24 LP; $16750, 18 LP.
▪ Lowest adult cost for outpatient visits in 5-17 year olds, $95, 1 LP.
▪ Highest adult cost for hospitalization in high-risk adults aged 18-49, $47722, 21 LP.
o Indirect costs of death – 2 methods: value of a statistical life (used to calculate total economic burden, includes intrinsic value of a human life) and present value of lost earnings (used to calculate total medical costs plus lost earnings).
- Probabilistic sensitivity analysis with Monte Carlo simulations and imputed parameter distributions.
- Estimated 24.7 million cases with 31.4 million attributed outpatient visits based on 2003 demographics. 334185 hospitalizations and 3.1 million hospitalized days. 44.0 million days of productivity lost. 41008 deaths amounting to 610656 life years lost.
- Economic burden of influenza
o Medical costs: $10.4 billion (10411.6 [4083.3, 22197.1]) (12% of total economic burden, below). 52% due to hospitalized cases, 30% due to outpatient cases, 18% attributed to treatment of patients who died from disease.
o Medical costs plus lost earnings: $26.8 billion (26757.2, [12813.0, 53155.9]), PVLE method of valuing lost lives.
o Total economic burden: $87.1 billion (87-67.3 [47215.3, 149508.6]), VSL method of valuing lost lives.
▪ 83% of this due to cases with deaths, 7% due to cases with hospitalizations, 8% due to outpatient cases. Medically unattended cases contribute only 2%.
▪ Lost earnings due to lost productivity and loss of life contribute 20%, $16.3 billion – $6.2 billion of which was due to lost productivity alone.
- Most costs – 64% of total costs – borne by those aged 65+.
- Annual burden per capita: Range $92 (only lost earnings) to $299 (including lost lives)
Schanzer, 2008
- Objective – Identify cohorts based on age, health status, and place of residence that accounted for the largest influenza mortality burden.
- Weekly death series with separate models fitted for each cohort. Poisson regression framework as before. Normalized influenza certified admissions useful as an influenza activity proxy because metric combines prevalence with severity. Regression coefficients generally as in Schanzer, 2008.
- Mortality rates for each cohort – denominator required, pulled from Statistics Canada data. Data on place of death combined with data on age-specific proportion of influenza-attributed hospitalization rates (Schanzer, 2008, previous work) used to determine fatality rates for hospitalized cases.
- Clinical attack rate of 5-10% assumed (see references 19-21).
- Time period: 1994-2000.
- Deaths – excess due to influenza
o 4000 deaths
o 14 per 100 000 population
o Increased from 23/100000 for those aged 65-69 years, to 831/100000 for those aged >= 90 years.
- Hospitalized case fatality rate due to influenza increased from 4% (age 50-64) to 30% (aged 90 years or older) – approximately similar to the hospitalized case fatality rates calculated from Molinari et al..
Background information for diabetes
Kannel, 1979
- Framingham study: Cohort of 5209 men and women aged 30 to 62 years in a small town served by a single hospital and a handful of physicians. Clinical cardiovascular endpoints diagnosed from biennial examinations and from hospital admissions, medical examiner’s reports, and other sources. Biennial examinations were also opportunities to reclassify patients on covariates (Markov assumption assumed in analysis, patient risk-time chopped into biennials). Mortality LTFU only 3%. Each biennial exam had 85% participation, with 69% total population participating over all examinations up to and including the tenth biennial examination or until time of death. Cohort initiated in 1949.
- Adjusted RRs for diabetes lower than crude RRs. Adjsuted for age, SBP, cigaretts per day, cholesterol, and LVH-ECG.
o CVD: Men 2.11 Women 2.03
o CHF: Men 1.82 Women 3.75
o Intermittent claudication: Men 4.16 Women 4.99
o Atherothrombotic brain infarction: Men 2.18 Women 2.17
o Coronary heart disease: Men 1.66 Women 2.06
o Cardiovascular disease death: Men 1.7 Women 3.3.
- Unadjusted rates ranged from 2.48 (CVD) to 4.72 (intermittent claudication) for men, and 3.57 (coronary heart disease) to 8.987 (intermittent claudication) for women. Presumably, all estimates statistically significant.
- PAR calculated to highlight public health importance of diabetes – contributes 7/1% of CVD death in men and 18/3% of CVD death in women; 5.0% of CVD in men and 7.3% of CVD in women (because PAR depends on absolute rates and prevalence of diabetes, these numbers should not be quoted any more.)
Brancati, 1997
- Measure the RR of ESRD related to diabetes mellitus in a large cohort of men from across the US.
- Multiple Risk Factor Intervention Trial (MRFIT), multicenter, RCT. Recruitment between 1973 and 1975. 12866 men randomized. (Other details – need further references).
- Primary outcome of present analysis = all-cause ESRD. Age-adjusted incidence of all-cause ESRD more than 12 times greater in diabetic than nondiabetic men, RR = 12.7 [10.5, 15.4], rates = 199.0 vs 13.7 cases per 100000 person-years.
- Fully adjusted model included race, age, SBP, cholesterol, income, smoking, history of MI. diabetes RR = 9.0 [7.4, 11.0]. Diabetes much stronger predictor (effect modification) in men aged 35 to 39 at baseline (RR 28.1 [16.3, 48.3]). Lowest for men aged 55 to 57 years at baseline (RR 6.7 [4.3, 10.4]). All models fully adjusted.
- Diabetes is a powerful risk factor for the development of ESRD.
- (Combine with prevalence statistics from ADSS).
- (Combine with information on high cost of ESRF – “affects only 0.07% of the population but consumes 2 to 3% of the health care budget” – Cite ADSS Atlas.)
Gu, 1998
- Estimate total and cause-specific mortality in diabetic and non-diabetic patients according to age, sex, race/ethnicity, and severity of diabetes.
- NHANES I baseline cohort interviewed 1971-1975, with follow-up to 1992-1993 (22 years). Data collected on sex, age, race, ethnicity, education, duration of diabetes, diabetes therapy, previous CVD, physical activity, height, weight, BP, total cholesterol. Analysis by age-standardization (direct method, using 1980 US population), by Kaplan-Meier analysis, and by CPH models (CPH results not reported, focus is on age-standardized mortality effect of diabetes including comorbidities).
- 13830 subjects, 5.1% with diabetes accounting for 10.6% of the deaths. Rates of mortality higher for men than women for both diabetic and non-diabetic patients, and increased with age. RR (diabetes vs no diabetes) closer to unity as age increased, and higher for women than men in the middle-elderly age groups). RR (diabetes vs no diabetes) similar for ethnic groups.
- All subjects
o Aged 25-44, diabetic rate = 12.4/1000 person-years, RR = 3.6, p 50 years).
- Twice as many visits to family physicians, 3 to 4 times as many visits to specialists (adults aged 20 to 49). Visit ratios decreased with age but for family physicians and specialists was about 1.5 for adults older than 50 years.
- Overall, mortality rates were twice as high for adults with diabetes (Adults = age 20 or older). Mortality rates were 4 to 6 times higher for adults aged 20-44, and 2 to 3 times higher for adults aged 45 to 79. In the 20 to 39 year age group, women with diabetes had, on average, a 9-year reduction in life expectancy. For men, this was 8 years.
Background information for diabetes and infection
Alexiewicz JM, Kumar D, Smogorzewski M, Klin M, Massry SG. Polymorphonuclear leukocytes in non-insulin-dependent diabetes mellitus: abnormatlities in metabolism and function. Annals of Internal Medicine, 123(12): 1995
- Blood samples taken from controls and patients with NIDDM at baseline, and after treatment with glyburide.
- ATP content of PMNs in patients with NIDDM significantly lower than that of controls.
- Phagocytic ability also significantly decreased.
- Significant improvement in phagocytic ability after treatment with glyburide.
- Phagocytosis measured with rate of ingestion of E. coli LPS coated oil droplets.
Delamaire M, Maugendre D, Moreno M, Le Goff MC, Allannic H, Genetet B. Impaired leucocyte functions in diabetic patients. Diabet Med, 14(1): 1997.
- Cross-sectional study of select PMN functions from the blood of 61 diabetic patients
o Type 1 – 40 patients
o Type 2 – 21 patients
- Baseline – PMNs from diabetic patients found to be more adherent to nylon, and also to express more adhesion molecules (CD11b and CD11c increased, CD11a decreased). Phagocytic properties not altered. Higher oxidative activity (bacteriocidal), measured by Nitroblue tetrazolium reduction and chemiluminescence.
- Upon stimulation, adherence and adhesion molecule expression similar between the two populations. No significant difference in number of NBT reducing cells. Chemotaxis and chemiluminescence decreased.
- These data are suggestive of a role for PMN altered function in the pathophysiology of atherosclerosis and infection in diabetic patients.
Gu K, Cowie CC, Harris MI. Mortality in adults with and without diabetes in a national cohort of the U.S. population, 1971-1933. Diabetes Care, 21(7): 1998.
- Objective – To estimate total and cause-specific mortality in diabetic and non-diabetic patients according to age, sex, race/ethnicity, and severity of diabetes.
- Prospective cohort study
o NHANES I baseline cohort interviewed 1971-1975, with follow-up to 1992-1993 (22 years) by follow-up surveys and vital statistics.
o Data collected on sex, age, race, ethnicity, education, duration of diabetes, diabetes therapy, previous CVD, physical activity, height, weight, BP, total cholesterol.
- Analysis
o Age-standardization – direct method, using 1980 US population
o Kaplan-Meier analysis
o CPH models
o CPH results not reported, focus is on age-standardized mortality effect of diabetes including comorbidities.
- 13830 subjects, 5.1% with diabetes accounting for 10.6% of the deaths. Rates of mortality higher for men than women for both diabetic and non-diabetic patients, and increased with age. RR (diabetes vs no diabetes) closer to unity as age increased, and higher for women than men in the middle-elderly age groups. RR (diabetes vs no diabetes) similar for non-Hispanic whites and blacks.
- All subjects
o Aged 25-44, diabetic rate = 12.4/1000 person-years, RR = 3.6, p ................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related download
Related searches
- words not found in dictionary
- zsh command not found python3
- file not found python
- npm command not found linux
- command pip not found ubuntu
- pip3 not found windows
- suspected condition not found icd 10
- concern not found icd 10
- pip not found linux
- university of alberta directory
- pycharm module not found error
- react router not found route