4. Global Burden of Tuberculosis
4
2 billion people are estimated to be infected
with Mycobacterium tuberculosis (TB).
1/3
In 2014, estimated 9.6 million new cases of TB,
1.5 million people died from TB.
Reference: World Health Organization: Global Tuberculosis Report.
In Book Global Tuberculosis Report City: World Health Organization; 2015.
6. Risk Factors of TB
• Age
• Male gender
• Low socioeconomic
status
• Malnutrition
• Substance abuse
• Silicosis
• Human
immunodeficiency virus
infection (HIV infection)
• Malignancy
• Diabetes
• Relative risk or hazard
ratio: 1.6 to 6.8
• Renal disease
• Celiac disease
• Gastrectomy
• Transplant
• Corticosteroids
• Tumor necrosis factor
inhibitors
6Reference: C Robert Horsburgh Jr. MD, MUS. Epidemiology of turberculosis. In: UpToDate, Elinor, LB MD (Ed),
UpToDate, Waltham, MA, 2013
7. Epidemiology of Diabetes Mellitus (DM)
7
422 million adults had diabetes, Prevalence Rate: 8.5 %
8.4 % in W Pacific
8.3 % in the Americas 7.1 % in Africa 8.6% in South-East Asia
7.3 % in Europe
Reference: World Health Organization: Global Report on Diabetes; 2016.
10 % in Taiwan,
2015
8. DM control vs. TB-1
The evidence is lacking
Reference: CC Leung et al, American Journal of Epidemiology 2008, 167: 1486-1494
HbA1C >=7% vs. no DM:
Active TB: RR 1.97 (95% CI 1.51-2.57)
9. DM control vs. TB-2
No evidence for any association between TB and dysglycemia
In the low TB-burden country of Denmark
Reference: A Leegaard et al, Diabetes Care 34:2530–2535, 2011
10. Taiwan’s National Health Insurance
Research Databases
Strength
• Large sample size
– 97% of Taiwan’s population
• Relatively inexpensive
• Real-world practice
– Medical service utilization
– Prescription drug use
• Longitudinal histories
Weakness
• Over-the-counter drugs?
• A secondary database
• Lag time
• Disease severity?
• Laboratory data?
10
Ref: Journal of Food and Drug Analysis, Vol 15, No. 2, 2007, Pages 99-108
12. Purpose
• To investigate the impact of DM and its
control on the risk of developing active TB
– The National Health Insurance Research
Databases(NHIRD)
– Time-dependent Cox proportional hazards models
12
13. Materials and Methods
Approved by the Research Ethics
Committee (REC) of National Taiwan
University Hospital
(NTUH REC: 201112111RIC)
13
14. Data Source
14
the Longitudinal Health Insurance Database (LHID) 2005
From 1996 to 2007
DM vs. non-DM
(case-control study,
Age, sex and time of entry,
1:1 in case number)
DM and co-morbidities
The risk factors for developing TB
Among DM
A Time-dependent approach
DM medications adherence
(weakness: no lab data (HbA1c) )
All selected cases were followed up until active TB developed,
31 Dec. 2007 or lost to follow up
15. Exclusion Criteria
15
Index date
: the date of first DM diagnosis
1 Jan 1997
Excluded:
Pts with DM and index date prior 1 Jan 1997
To ensure an observation period of 1 yr
Excluded:
< 6 months of follow up after index date
To ensure a sufficient observation period
Pts with TB prior to the index date were also excluded
Higher risk for getting another episode of TB
16. Definition of DM or TB
16
ICD9: 250
A code: A181
Exclude gestational diabetes
ICD9: 010-018
A code: A020,A021
Anti-TB medications:
Isoniazid, ethambutol, rifampicin, pyrazinamide…etc.
17. Co-morbidities
• Malignancy
• End-stage renal disease (ESRD)
• Chronic obstructive pulmonary disease
(COPD)
• Pneumoconiosis
• Liver cirrhosis
• Autoimmune diseases
• Acquired immunodeficiency disese
• The low income group
– Annual household income < 4,500 US dollars 17
Ref: PLoS One 2012; 7: e37978.
18. Statistical Analysis-1
• Part 1 (case control)
– Inter-group differences
• Numerical variables: independent samples t-test
• Categorical variables: chi-square or Fisher’s exact test
– Curves of time-to-active TB
• Kaplan-Meier method and the log-rank test
– Factors for developing TB
• Cox-regression analysis
– Sensitivity analysis
• In different sub-population with different follow-up
duration 18
19. Statistical Analysis-2
• Part 2 (Among DM patients)
– Factors for developing TB
• The time-dependent Cox proportional hazards model
– Time-dependent variables:
» Age, co-morbidities, use of systemic corticosteroids, oral
hypoglycemic agents (OHAs), and insulin,
» Adherence to anti-DM medications, DM associated
admissions
– Time-independent variables: sex
– Sensitivity analysis
• DM pts need > 90 days of DM medications
• All analyses were performed by using the SAS
19
20. Factors During Time Segment
20
TB event date-90 days-270 days
• OHAs: defined daily doses (DDDs)
• Adherence to anti-DM medication, continuous variable, 0 to 1
• The proportion of days covered by anti-DM medication prescribed at the
outpatient clinic within 270-day time segment
• The average OHA daily dose
• The number of days covered by systemic corticosteroids
• The maximum average daily dose (MADD)
• The maximum of the average doses of OHAs and insulin for every 90
consecutive days
• Insulin use
• Insulin dose during admission and the outpatient clinic (continuous variable)
• Insulin use during admission and at the outpatient clinic (categorical variable)
• The number of admissions with compatible DM diagnoses
21. Delayed Diagnosis of TB and
Reactive Hyerglycemia
• Two sensitivity analyses
– Risk factors for developing of TB > one years after
the diagnosis of DM
– The 270-day time segment
• 360 to 90 days, 450 to 180 days prior to each outcome
event
21
30. DM cases with or without TB
30
Age 63.6±13.2 yrs
Male dominance
31. DM patients who developed TB received higher
doses of insulin, OHAs, systemic corticosteroids
31
32. Time-dependent Cox proportional
hazards analysis
32
Sensitivity analysis
•Sensitivity analyses revealed the same results
1. developing TB more than 1 year after the diagnosis of DM
2. the 270-day time segment to from 450 days to 180 days prior to each outcome
33. Discussion
• Three major findings
– DM is an important risk factor of TB, and its effect
persists for at least 5 yrs
– The risk of TB parallels the severity of DM
• Measured by the number of DM-related admissions,
MADD of OHA and insulin use during admission
– Some cases of TB may be prevented by promoting
adherence to anti-DM medication
33
34. Increased the Risk of TB in DM
Patients
• Risk in previous studies: 1.48 to 6.8
– Our study: 1.293 (1.154-1.449)
– Taiwan’s 2001 National Health Interview Survey:
2.09 (1.10-3.95)
• Pros: including environmental, educational and socio-
economic conditions
• Cons: no autoimmune disease, malignancy, ESRD, liver
cirrhosis
• Mechanisms
– The compromised immune response in diabetic pts
34
Ref: 1. Lancet Infect Dis 2009; 9: 737–746. 2. Clin Infect Dis 2012; 54: 818–825. 3. Clin Infect Dis 2008; 47: 634–641
35. OHA and insulin
• Previous studies:
– Insulin dependence is a marker of disease severity
and predictive of increased risk of TB
• The daily dose of OHA and insulin continued
to increase after DM was diagnosed.
• Higher daily dose of OHA
– Worse therapeutic response and longer duration
of DM
– Increase the risk of developing TB
35
Ref: Am Rev Tuberc 1952; 65: 1–50.. Trop Doct 1990; 20: 147–150.
36. DM-associated Admissions and Insulin
• Higher HbA1c: higher risk of admissions
• Insulin use during admission
– A surrogate maker of poorly controlled DM
• DM-associated admissions
– Diabetic ketoacidosis and non-ketotic hyper-
osmolar syndrome
– Poor DM control
– Increase the risk of contact with TB cases
36Ref: 1. Arch Intern Med 1999; 159: 2053–2057. 2. Arch Intern Med 1997; 157: 669–675. 3. Clin Chest Med 1989;
10: 397–405.
37. Insulin Dose during Admission or
the Outpatient Clinic
• Those are not independent risk factors
– Higher dose reflects a more aggressive use of
insulin
– A higher probability of good sugar control
37
38. Delayed TB Diagnosis
• Similar final statistical models:
– Sensitivity analysis of developing TB > one years
after diagnosis of DM
– Adjusting the 270-day time segment prior to each
outcome event
• Suggest that delayed TB diagnosis and
reactive hyperglycemia
– NOT alter the impact of DM and its control on the
risk of TB
38
39. The Protective Effect of Adherence
to Anti-DM Medications
• Even greater in DM pts frequently requiring
anti-DM medication
• Poor adherence to anti-DM treatment
– Poor diabetes control
– Increased susceptibility to infection
• Similar results in Hong Kong
– HbA1c > 7 %
39
Ref: 1. Clin Ther 2011; 33: 74–109 . 2. Diabetes Care 2008; 31: 916–921 3. QJM 2007;
100: 345–350 4. Am J Epidemiol 2008; 167: 1486–1494.
40. Other Risk Factors
• Age: lower HR than that of Baker’s study
– The prevalence of underlying co-morbidity and
the number of DM-associated admissions often
increased as age increases
• Male sex, COPD, DM, ESRD and the use of
systemic corticosteroids are also well-known
predisposing factors of TB
– Smoking (not available in NHIRD), TB and COPD
40
Ref: 1. Clin Infect Dis 2012; 54: 818–825. 2. Am J Respir Crit Care Med 2009; 180: 475–480. 3. Am J
Respir Crit Care Med 2007; 176: 532–555 4. Semin Dial 2003; 16: 38–44.
41. Limitations
• Lack of culture and laboratory data for the
diagnosis of TB
– Diagnosis of TB has been verified
– Sensitivity analysis
• Without lifestyle information
– DM patients may quit smoking and drinking after
diagnosis
– It may reduce the risk of TB
• Using adherence, not HbA1c to measure DM
control
– Prospective clinical studies 41
Ref: PLoS One 2012; 7: e37978.
42. Conclusions
• This study confirms the association of DM and
TB and complements previous reports by
showing that the risk of developing TB
parallels the severity of DM.
• Some cases of TB can be prevented by
promoting adherence to anti-DM medication.
• Healthcare providers should keep a high index
of suspicion and periodically screen for active
TB in DM patients.
42