By Aman Narayan
MD Candidate, UT Southwestern Medical School
Rice University ’18
As the COVID-19 pandemic continues despite increasingly widespread vaccine distribution, it is no secret that our public health response has been disastrous and misguided at every level. One tool in the public health arsenal is contact tracing, the process of identifying people who may have had close contact with an infected individual. When rigorously conducted, contact tracing can shed light on numerous epidemiological characteristics of a disease. Unfortunately, no centralized contact tracing effort has been established, leaving individual states and counties to collect data for themselves. We aimed to test whether differences in public funding across states could explain regional variation in the quality of case report data, which is the foundation of contact tracing.
There is no standardized survey consistently administered among different contact tracing programs. In most states, individual counties report to the state public health agency, which then reports case information to the Centers for Disease Control (CDC). A previous Baker Institute study highlighted disparities in the completeness of the CDC case report data across states. The study first looked at the quality and accuracy measures of 42 states for which the CDC had any data. The comprehensiveness of the CDC case reports was assessed by comparing case counts to data reported by the New York Times (NYT), which was widely seen as a more accurate measure of the true number of COVID-19 cases. For the period May 5 – July 19, 2020, more than half the states (22) had CDC to NYT case report ratios less than 80%, while 10 states had ratios less than 50%. The state with the lowest ratio was Texas: 5.5%. The study also found that while 40 states had at least some percentage of their case reports include racial/ethnic background data on the individual, many states (24) did not report data on exposure history on any of their case reports. The lack of complete information about exposure history could represent one barrier to identifying transmission prevalence or risk among certain activities or places within the community.
To compare the completeness of state case report data to state-level public health spending, we used data from the 2019 Association of State and Territorial Health Officials (ASTHO) Profile Survey data set, which relies on self-reported information from state public health agencies. We examined 2019 data on overall public health spending as well as state spending on these subcategories: health preparedness, infectious diseases, health data and health laboratories. State population data was obtained using 2017 Census Public Use Files.
The median total public health expenditure per capita was $82. Among states that reported data, the average number of case reports including exposure history was 14% and the average number of case reports including racial/ethnic background was 62%. North Dakota was the state with the highest per capita public health spending at $2,403. Fifty-three percent of its case reports included data on exposure history, but despite its high spending, none included data on racial/ethnic background. Massachusetts was a state with a relatively low per capita public health spending of $15. Seventy percent of its case reports included data on racial/ethnic background and 42% included data on exposure history. Additionally, the state’s CDC to NYT case report ratio was 88%, suggestive of a fairly accurate case count within the state.
We applied regression analysis to test whether various measures of public health spending (on infectious diseases, hazard preparedness and health data and overall state public health expenditures) were associated with the percentage of state case reports with non-missing data on racial/ethnic background. Paradoxically, we found each additional dollar per capita spent on health data was associated with a 10% decrease in the state’s case reports with non-missing data on racial/ethnic background (p=0.01; where a p-value less than 0.05 would be significant at the 95% confidence level). The rest of these measures of spending were not statistically significant, with p-values ranging from 0.46 to 0.78. Similar regression analyses revealed that the state’s percentage of case reports that included data on exposure history was not associated with any change/level in public health spending, with statistically insignificant p-values ranging from (p=0.16 to 0.47). Further analysis demonstrated that the percentage of case reports that included data on exposure history was also not associated with any level of spending on any of the subcategories.
We then compared these case report quality measures to self-reported full-time equivalents (FTEs) per capita in the state public health department. Again, the regression analyses suggested no statistically significant association between public health department FTEs and the percent of case reports that included data on racial/ethnic background or exposure history (p=0.71, p=0.28, respectively). We found a wide variation of per capita state public health spending among the states whose CDC to NYT case report ratios were close to 1, suggesting that public health spending did not affect the case count accuracy reported to the CDC. Each percentage increase in the CDC to NYT case report ratio among those states whose ratios were less than or equal to 1 was associated with 3 per 10,000 additional FTEs per capita. However, this relationship was not statistically significant (p = 0.20).
Altogether, the data we examined did not identify meaningful predictors of contact tracing quality, nor accuracy of case reporting. Thus, it is difficult to determine whether increased public health spending or more public health FTEs would contribute to improved contact tracing quality. These findings may be attributed to the lack of centralized data collection efforts within states; in other words, public health funding and staffing on the state level may not have had any real influence on contact tracing programs at the county level. It is also possible that stronger guidance at the federal level or concerted cooperation between the federal government and state governors is a prerequisite for greater public health funding to be effective. Further analysis would need to be conducted at the county level, possibly with more specific data on pandemic-related expenditures, to identify meaningful predictors of quality and accuracy in case reporting. Our analyses suggest that states’ funding of public health work may be a necessary criteria for conducting accurate case reporting in the midst of a pandemic, but it is not sufficient.