Basic Investigation of Outbreaks

1
Basic Investigation of Outbreaks

• Karin Galil, MD MPH
• Centers for Disease Control and Prevention
• Atlanta, Georgia

2
Outline

• Identify the outbreak
• Investigate the outbreak
• Interpret results
• Institute control measures
• Report results

3
Identify Potential Outbreaks

• What is an outbreak ?
• How can one detect outbreaks ?
• Why should one look for outbreaks ?

4
Outbreak Definition

• An increase in the occurrence of a complication
  or disease above the background rate
• One rare event
• e.g. GAS surgical site infection
• Many episodes of common occurrence
• e.g. MRSA surgical site infections

5
Background Rate of Disease

• Ongoing surveillance
• Determine ratescompare within and between
  institutions
• Trends
• Requires common, accepted case definitions
• Retrospective review of data

6
Pitfalls in Rate Estimates

• Case definitions
• Numerator
• Different definition? increased or decreased
  number
• Population at risk
• Denominator
• Different definition? increased or decreased rate

7
Who Identifies Potential Outbreaks ?

• Routine surveillance
• Infection control
• Registries
• Clinical staff
• Laboratory staff

8
Reasons to Investigate

• Outbreak control
• Increased knowledge
• Pathogen
• Risk factors for acquisition
• Transmission
• Epidemiology

9
Clusters that Suggest Nosocomial Transmission

• Similar cases on one unit or among similar
  patients
• Cases associated with invasive device
• HCW and patients with same infection
• Typical nosocomial pathogen
• multiply-resistant
• opportunistic

10
Determining Risk Factors for Disease

• Known risk factors in hospital-acquired
  infections
• Invasive devices
• Severe illness or underlying disease
• Environmental factors
• Especially immunocompromised patients (e.g.
  aspergillosis)

11
Institute Control Measures

• Immediate control measures needed even before
  investigation begun or completed
• Simple e.g. improved handwashing
• Complex cohorting patients, closing unit,
  halting use of device or product

12
Before the Investigation

• Cooperation
• All involved personnel and administration
• Laboratory capacity
• Antimicrobial susceptibility testing, typing
  (molecular and nonmolecular methods)
• Resources
• Personnel, supplies, lead investigator,
  statistician

13
The Investigation

• Define case
• Find cases
• Confirm outbreak
• Review charts
• Describe epidemiology
• Generate hypothesis
• Test hypothesis
• Analyze data
• Communicate results

14
Case Definitions

• Working case definition
• Person, place, time
• Clinical, laboratory or diagnostic findings
• Confirmed vs. possible cases
• Case definitions usually change during the
  investigation

15
Example Case Definition

• A case of multi-drug resistant tuberculosis
  was defined as any patient in Hospital X
  diagnosed with active tuberculosis from January
  1, 1999
• to December 31, 1999 whose isolate was resistant
  to at least isoniazid and rifampin.

16
Case Finding

• Use case definition to find other cases in the
  source population
• Large potential source population discharge
  diagnoses, microbiology log books, emergency room
  visits, use of diagnostic technique
• Small population (unit of hospital) review
  charts of entire cohort

17
Line Listing
18
Confirm the Outbreak

• Calculate background rate of disease
• Compare rate during outbreak with background rate
• Define periods from incubation timeto last case
  (or present)

19
Rate Ratioattack rate (outbreak
period)attack rate (background period)
20
Pseudo-Outbreaks

• Clusters of positive cultures in patients without
  evidence of disease
• Perceived increase in infections
• New or enhanced surveillance
• Different laboratory methods

21
Descriptive Epidemiology

• Line listing of case-patients (person, place,
  time)
• Demographic information
• Clinical information
• Epidemic curve
• Point source
• Person-to-person

22
Point Source Outbreak

• Shorter duration
• Sharp peak in epidemic curve
• Rapid resolution
• May resolve without intervention

23
Epidemic CurvePoint Source Outbreak
24
Epidemic CurveContaminated Product
N87
Number of persons with abscess
1995
1996
25
Bloodstream Infections and Pyrogenic
ReactionsExtrinsic Contamination
26
Person-to-Person or Contaminated Equipment

• Poor infection control technique or contaminated
  patient equipment
• Long duration
• May not resolve without intervention
• If HCW and patients affected, plot separately and
  together to determine mode of transmission

27
Clues

• Location
• Tb skin test conversion associated with
  outpatient HIV clinic?air flow
• Patient characteristics
• Immunocompromised patients
• Persons of a certain age
• Persons with same disease/procedure

28
Hypotheses

• What caused the outbreak ?
• Available data from the outbreak
• Published literature
• Expert opinion
• Hypothesis testing

29
Epidemiologic Studies

• Case-control studies
• Cases disease
• Controls equal likelihood of exposure as cases
• Cohort studies
• Cohort selected on the basis of exposure status

30
Case-Control Study

• Advantages small number of cases, better for
  rare diseases, diseases with long latency
  periods, multiple exposures
• Disadvantages selection and recall bias, not
  good if exposure is rare, cannot measure disease
  incidence rate (OR vs. RR)

31
Cohort Study

• Advantages can study rare exposures, can
  calculate disease incidence rates, selection bias
  less likely
• Disadvantages feasibility, not suited to rare
  diseases

32
Collect Data

• Complete same data for cases and controls
• Unbiased same way to avoid bias

33
Potential Types of Bias

• Selection bias
• Self-selection
• Diagnostic bias
• Information bias
• Differential vs. misclassification
• Recall bias

34
Questionnaire

• Design questionnaire
• Demographic information
• Potential risk factors
• Outcomes
• Field test
• Complete for on all patients

35
Enter and Clean Data

• Line listing
• Statistical program
• EpiInfo, SAS, STATA
• Clean data
• Correct errors

36
Data Analysis

• Descriptive statistics
• Univariate analysis
• Stratified analysis
• Complex analysis

37
Descriptive Statistics

• Vital first step
• Describe person, place, time
• Describe frequency of all variables collected
• Look for errors
• Decide on further analysesbased on these results

38
.
Disease
Yes
No
Yes
Exposure
No
39
Risk Estimate

• OR/RR gtgt 1
• Strong positive association
• OR/RR 1
• No association
• OR/RR ltlt 1
• Strong negative association

40
Statistical Significance

• Confidence Intervals
• Include 1
• Exclude 1
• P value
• p gt 0.05
• p ltlt 0.05

41
Univariate AnalysisCategorical Variables

• Categorical variables (yes/no young/old)
• Odds Ratio (OR) ? case-control study
• Relative Risk (RR) ? cohort study

42
Odds Ratio

• Case-control study
• OR odds that person with disease was exposed
  compared to odds that a person without disease
  was not exposed to risk factor
• OR estimates the relative risk

43
Odds RatioOR ad / bc
44
Odds Ratio
45
Calculating the Odds RatioOR ad /
bcOR (14)(8) / (7)(5)OR 3.2
46
Relative Risk

• Cohort study
• RR risk ratio incidence rate ratio relative
  rate
• RR risk of disease among exposed compared to
  risk among the unexposed

47
Relative RiskRR a(cd) / c(ab)
48
Confidence Intervals

• Sampling ? estimates the OR or RR
• 95 confidence Intervalsif we resampled numerous
  times, our estimate would fall within these
  bounds 95 of the time
• Finite population correction

49
Statistical Tests for 2x2 Tables

• Chi-square test
• Fishers exact testif value of any cell lt5
• P value indicates level of certainty that
  association was not due to chance alone

50
Risk Estimate vs. P Value

• OR or RR direction strength of association
• gtgt1 strong association
• 1 no association
• ltlt1 strong inverse association
• P Valuelevel of certainty about the estimate of
  the association
• ltlt.05 unlikely to be due to chance

51
Univariate Analysis Continuous Variables

• Continuous variables (e.g. age, bp)
• Distribution
• Normal (bell-shaped)
• Mean and standard deviation
• Not normal
• Median and range

52
Stratified Analysis

• Simple stratified analysis
• Control for one variable
• Logistic/linear regression models
• Control for multiple variables at once
• Control for confounding and effect modification
• Non-linear relationships

53
Microbiologic Investigation

• Alert lab save all specimens positive cultures
• Typing of organisms
• Species identification
• Biotyping
• Antimicrobial susceptibility testing
• Advanced typing (serotyping, plasmid analysis,
  phage typing, isoenzyme electrophoresis, genetic
  fingerprinting)

54
Environmental Investigation

• Are inanimate objects linked with the outbreak ?
• Were infections clustered in one area ?
• Consider infected devices, medications/products,
  airflow patterns

55
Interpret Results

• Is there an association ?
• It is statistically significant ?
• Was study biased ?
• Are the results plausible ?
• Did the exposure precede the outcome ?
• Are results consistent with other studies ?
• Is there a dose-response effect ?

56
Control the Outbreak

• Routine infection control procedures
• Guidelines for universal precautions
• Specific guidelines for patient-care equipment
• Specific interventions for the ongoing outbreak
• Cluesperson, place, time

57
Evaluate Control Measures

• Did the control measures stop
• the outbreak?
• Were there multiple modes of transmission ?
• Were control measures implemented properly ?
• Were control measures sufficient ?

58
Implement Successful Control Measures
59
Report Results

• Inform all concerned parties of results
• Hospital staff, consultants, health department
• Contaminated products/devicesgovernment
  authorities, manufacturers
• Media spokesperson

60

• Investigations are
• Challenging
• Time - consuming
• Imperfect