Every year, conflicts and natural disasters affect millions of people worldwide. However, the resources to assist those affected are perpetually insufficient. When emergencies strike, assistance organizations must decide where to prioritize their limited resources to reduce as much mortality and suffering as possible.
At the start of their emergency response activities, organizations typically make three key decisions: 1) determine if they will respond 2) prioritize/triage the needs of the affected population; and 3) choose first response programs to implement. Many studies and authors note that these decisions are often based on insufficient evidence and personal judgement.
This dissertation argues that we, as emergency responders, can do better. Epidemiologic methods can empower us to make better decisions based on better measurement, analysis, and evidence – improving outcomes for emergency affected persons, globally. This dissertation provides three examples of epidemiologic methods being used to inform critical emergency response decision points.
Aim 1 addresses the first emergency response decision: prioritizing the most severe emergencies for response programing. Aim 2 focuses on the second emergency response decision: how can responders most accurately prioritize the needs of affected persons by using needs surveys, given the potential that needs vary by gender or age. Aim 3 examines the third decision: which response programs to implement, by summarizing the evidence base for the effectiveness of standard emergency programs.
Methods
In aim 1 I facilitated a panel of outbreak specialists from a leading emergency response organization to develop, test, and validate a new measure for the classification of outbreaks. I used classical scale development methods, including both qualitative and quantitative procedures.
In aim 2 I used data from 12 emergency needs surveys to examine a common assumption that reported needs and experiences vary based on the gender and/or age of the respondent. I conducted both individual analyses of each study as well as a set of meta-analyses examining the prevalence differences found between gender and age sub-groups.
In aim 3 I conducted a systematic scoping review of the evidence of what programs are effective in acute emergency settings. I searched six academic databases as well as eight sector-relevant grey literature databases -focusing on evidence for standard emergency interventions.
Results
In aim 1, a new outbreak classification measure was successfully developed based on inputs from the expert panel and a compiled dataset of indicators in global outbreak emergencies. The measure allows for the immediate (within two hours) classification of outbreaks. The expert panel participated in qualitative exercises where they developed a construct of ‘scale and severity of outbreak emergencies.’ This construct had four sub-dimensions, and a scale was developed to measure each sub-dimension, and then combined into a single measure. The content validity, criterion validity, construct validity, and reliability were examined for the measure. Criterion validity was based on a strong (0.87) correlation between the new outbreak measure and a ‘gold standard’ ranking of outbreak emergencies created by a group of emergency decision-makers (‘judges’). Similarly, construct validity was based on the measure performing as predicted when compared to measures of a similar/dissimilar construct, (convergent and divergent validity). The case for reliability was made using intraclass correlations between the new outbreak measure and the ‘gold standard’ measure (a robust result of a 0.87 using an ICC 3, 1), as well as comparing how well the outbreak measure worked alongside the conflict and natural disaster measures (another robust finding of 0.91 using an ICC 3, 1).
In aim 2, I found that emergency affected persons of various gender or age groups very rarely differ in their responses to needs and experience questions in emergency surveys. When searching for differences in how gender or age groups report their households’ top three needs, meaningful differences in individual studies were found 6% of the time. When a meta-analysis of the same data was conducted across all needs questions in all 12 surveys, no meaningful differences were found between how either men or women report needs, or how different age groups report needs. Responses to questions about experiences (rather than needs) in emergencies were slightly more likely to vary by gender or age group. The meta-analysis of experience questions showed that across the 12 assessments differences in how gender or age groups experience emergencies were extremely rare (less than 4% of questions showed a meaningful summary prevalence difference).
In aim 3 I identified 43 programs that are commonly implemented in acute emergency response. My scoping review searched for any studies that rigorously evaluated the impact of one or more of these programs. My search identified 4,005 unique studies; I screened them all for eligibility, resulting in only four studies that met all inclusion criteria. Thus 39 of the pre-identified, common emergency programs have no published evidence of their effectiveness in acute emergencies. The remaining four, each have one study in one context that demonstrates at least one positive effect of the program.
Conclusion
This dissertation provided evidence that epidemiologic methods can help solve problems, answer questions, and improve the allocation of resources in acute emergencies. While each aim focused on a unique decision point within acute emergency responses, they all contended with similar difficulties, such as incomplete and poor-quality data and a lack of shared definitions for what data points are relevant in decision-making. Yet in all three aims I found other similarities as well: there are relevant data available; and there are effective methods available that can answer many of our questions.
Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/w6zv-ac51 |
Date | January 2024 |
Creators | Morris, Bobi Janelle |
Source Sets | Columbia University |
Language | English |
Detected Language | English |
Type | Theses |
Page generated in 0.0094 seconds