In an article recently published in BMC Public Health entitled Evaluating completion rates of COVID-19 contact tracing surveys in New York City, authors, including PRI Epidemiology and Modeling Co-Lead Jeffrey Shaman and PRI Faculty Member Sen Pei, identified factors associated with completion rates of COVID-19 contact tracing surveys in New York City (NYC) and evaluated the utility of a predictive model to improve completion rates. Researchers analyzed laboratory-confirmed and probable COVID-19 cases and their self-reported contacts in NYC from October 1st 2020 to May 10th 2021, which included 742,807 case investigation calls.
Key Findings
- Completion Rates and Variations: The completion rate of contact tracing surveys in NYC during the study period was 79.4%. There were substantial variations in completion rates across different ZIP code areas within the city.
- Factors Influencing Completion Rates:
- Age: The age of the index case significantly impacted the completion rate. Compared to young adults (age 24 to 65), seniors (age over 65) had a lower completion rate by 12.1%, and the completion rate for the youth group (age under 24) was lower by 1.6%.
- Time of Day of Phone Call: Phone calls made between 6 pm to 9 pm had a 4.1% higher completion rate compared to calls made between 12 pm to 3 pm.
- Demographic and Socioeconomic Factors: Variables such as the percentage of Black and Hispanic residents, median household income, and average household size in ZIP code areas were associated with completion rates.
- Predictive Model for Improving Completion Rates: Researchers developed a random forest model to predict the best time of day for phone calls to maximize completion rates. Age and time of day of the phone call were identified as the two most important variables in prediction. While the overall improvement in completion rates using the predictive model was marginal (1.2%), certain ZIP code areas showed improvements of up to 7.8%.
- Implications for Operations: The findings suggest potential strategies for optimizing completion rates, such as adjusting phone call schedules based on age groups and time of day. The study highlights the importance of tailoring contact tracing efforts to specific community demographics and behaviors to improve effectiveness.
Conclusion
The study acknowledges limitations such as the availability of only a few individual-level variables for analysis, the need for cautious interpretation of predictive model results, and potential factors like local demographics and cultural differences that could impact completion rates. However the research provides insights into the completion rates of COVID-19 contact tracing surveys in NYC and proposes strategies, including the use of predictive models, to enhance completion rates, particularly in specific demographic and geographic contexts.
Read the full article at BMC Public Health.