ADVANCEMENT OF THE SCIENCE
at the local level. This population included 1,095 observations of cycles from the Retail Program Standards that were completed by 770 individual LHDs (Table 2). Data and Analysis Using the listing from FDA (2023b) of juris- dictions enrolled in the Retail Program Stan- dards, we were able to observe the enroll- ment date and conformance with each of the standards in each 5-year SA cycle for all LHDs. Using these data, we created five out- come variables: 1) average time to SA sub- mission, 2) average time to VA following suc- cessful SA, 3) number of standards achieved via SA, 4) number of standards achieved via VA, and 5) number of SA updates. SA cycles were broken down into four groups: Cycle 1 (770 observations), Cycle 2 (253 obser- vations), Cycle 3 (64 observations), and Cycle 4 (8 observations). A dichotomous variable for participation in the Mentorship Program and a continuous variable for the size of population served by a jurisdiction were created using internal NACCHO data sources, including the National Profile of Local Health Departments . Grant funding for the Retail Program Standards through FDA for RPS CAP was identified using public funding data. Given that the data set used in our study contains the entire population of enrolled LHDs, 95% confidence intervals are not reported in text or tables beside their parameter estimates. Eect Size Calculations We used analysis of variance (ANOVA) to assess dierences in outcome variables across several groupings: Mentorship Program par- ticipants, RPS CAP grantees, SA cycles, and jurisdiction size. Due to the number of statis- tical comparisons made in our study, Cohen’s d and Cohen’s h were calculated for each eect as a measure of the magnitude of the eect in the comparisons of continuous and dichoto- mous outcomes, respectively, in lieu of p -val- ues. Eect size measures are a standardized metric to compare dierences between two means or proportions. In our analysis, we considered values of <0.2 as non-meaningful, ≥0.2 as small, ≥0.5 as medium, ≥0.8 as large, and ≥1.2 as very large, in line with previous interpretations of Cohen’s d and Cohen’s h (Cohen, 1988; Sawilowsky, 2009). We high- light comparisons with large and very large
TABLE 2
Demographic Information of Local Health Departments ( N = 770) Enrolled in the Voluntary National Retail Food Regulatory Program Standards
Demographic
# (%)
Grant program participation Mentorship Program
87 (11.3) 37 (4.8) 23 (3.0) 623 (80.9)
RPS CAP
Mentorship Program and RPS CAP
Neither
Jurisdiction population size * Small (<50,000)
199 (25.8) 309 (40.1)
Mid-sized (50,000–500,000)
Large (>500,000)
86 (11.2)
Missing
176 (22.9)
* Population size was determined using NACCHO member profiles. Missing jurisdictions did not have a current NACCHO profile. Note. RPS CAP = Retail Program Standards Cooperative Agreement Program; NACCHO = National Association of County and City Health Officials.
TABLE 3
Average Time to Submission of Self-Assessment (SA)
SA Submitted Past Deadline (Mean # of Months) Cycle 1 Cycle 2 Cycle 3 Cycle 4
Total
Grant participation
Mentorship Program
5.0
27.9
1.4
1.0
12.0
RPS CAP
-2.0 -1.5
3.1
-3.7
– –
0.3
Mentorship Program and RPS CAP
24.0 14.7
-14.3
10.2
Neither
6.1
2.0
-7.3
8.0
Jurisdiction population size * Small (<50,000)
3.3 6.7
0.1
-5.3
–
2.5 8.4
Mid-sized (50,000–500,000)
13.4 24.4 15.3
5.2 0.5 0.4
4.8
Large (>500,000)
10.5
–
14.0
Average
6.0
-5.3
8.0
* Population size was determined using NACCHO member profiles. Missing jurisdictions did not have a current NACCHO profile. Note. RPS CAP = Retail Program Standards Cooperative Agreement Program; NACCHO = National Association of County and City Health Officials.
Regression Models Negative binomial regression was used to estimate the relationship between participa- tion in the Mentorship Program and RPS CAP
eect sizes, as those independent variables are the most likely to be significantly asso- ciated with improved conformance with the Retail Program Standards.
18
Volume 86 • Number 4
Powered by FlippingBook