HOY Risk Factors For Salmonella Contamination of Whole Chicken Carcasses Following Changes in U.S. Regulatory Oversight

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

1713

Journal of Food Protection, Vol. 84, No. 10, 2021, Pages 1713–1721
https://doi.org/10.4315/JFP-21-144
Copyright Ó, International Association for Food Protection

Research Paper

Risk Factors for Salmonella Contamination of Whole Chicken


Carcasses following Changes in U.S. Regulatory Oversight
AARON T. E. BECZKIEWICZ https://orcid.org/0000-0002-0996-619X1 AND BARBARA B. KOWALCYK1,2*

1Department of Food Science and Technology and 2Translational Data Analytics Institute, The Ohio State University, Columbus, Ohio 43210, USA
MS 21-144: Received 2 April 2021/Accepted 25 May 2021/Published Online 28 May 2021

ABSTRACT
Salmonella is a common cause of foodborne illness in the United States and often is linked to chicken products. Salmonella
contamination has been associated with meat processing facility characteristics, such as the number of employees (i.e., hazard
analysis critical control point [HACCP]–based definition of size). The risk factors for Salmonella contamination in U.S. poultry
have not been evaluated since implementation of the New Poultry Inspection System (NPIS) in 2014. The goal of this study was
to determine whether risk factors for Salmonella contamination changed after implementation of the NPIS. Presence or absence
of Salmonella in whole chicken carcasses was modeled using microbiological testing data collected from 203 poultry processing
establishments by the U.S. Department of Agriculture Food Safety and Inspection Service between May 2015 and December
2019. A model was fit using generalized estimating equations for weekly presence or absence of Salmonella, and production
volume, geographic location, and season were included as potential covariates among other establishment demographics. Odds
ratios (ORs) and 95% confidence intervals (CIs) were calculated from the marginal model. Of the 40,497 analyzable samples,
1,725 (4.26%) were positive for Salmonella. Odds of contamination was lower among establishments slaughtering 10,000,000
birds per year (OR ¼ 0.466; 95% CI, [0.307, 0.710]) and establishments producing ready-to-eat finished products (OR ¼ 0.498;
95% CI, [0.298, 0.833]) and higher among establishments historically (previous 84 days) noncompliant with HACCP
regulations (OR ¼ 1.249; 95% CI, [1.071, 1.456]). Contamination also significantly varied by season and geographic region,
with higher odds of contamination during summer and outside the MidEast Central region. These results support continuation of
targeted food safety policies and initiatives promoting pathogen reduction by establishments with smaller volumes and those
noncompliant with HACCP regulations.

HIGHLIGHTS
 Low production volume and summer season are risk factors for Salmonella contamination.
 Salmonella contamination varied by region, even after adjusting for seasonality.
 Facilities producing ready-to-eat product have lower odds of contaminated raw product.
 HACCP noncompliance (9 CFR §417.4) was associated with Salmonella contamination.

Key words: Chicken; Hazard analysis critical control point; Inspection noncompliance; Risk factor; Salmonella

Salmonella is a common cause of foodborne illness in Identification of risk factors for Salmonella contami-
the United States, accounting for approximately 1.02 million nation of food is an important step for reducing disease
illnesses, 20,000 hospitalizations, and 400 deaths annually burden, and mitigating these risk factors will likely have the
(18). Three of the most common Salmonella serotypes greatest impact. Risk factor research often focuses on
associated with human illness (Enteritidis, Newport, and preharvest conditions (e.g., rearing conditions (8) and feed
Infantis) are frequently recovered from poultry. Approxi- withdrawal times (10)), but such mitigation strategies can
mately 196,000 illnesses are linked to chicken consumption be difficult to implement industry-wide due to the
annually, resulting in annual economic losses of $1.1 billion complexity of pathogen control in animal agriculture (1).
to $2.8 billion (2, 19, 20). These cost estimates do not Mandatory control of pathogens already exists in processing
include government or industry costs such as those facilities, and industry-wide mitigation strategies may be
associated with investigations, recalls, and legal action (17, easier to implement if they target specific risk factors.
26). Given the magnitude and scope of this Salmonella issue, Further, food safety practices are driven in large part by
targeted efforts regarding chicken and other poultry products processing facility characteristics, such as the number of
could significantly reduce the burden of salmonellosis. employees or the amount of product produced (4, 12).
In previous studies, the U.S. Department of Agriculture
* Author for correspondence. Tel: 614-292-6281; Fax: 614-292-0218; (USDA) Food Safety and Inspection Service (FSIS) has
E-mail: kowalcyk.1@osu.edu. evaluated the relationship between establishment character-
1714 BECZKIEWICZ AND KOWALCYK J. Food Prot., Vol. 84, No. 10

TABLE 1. Major changes in U.S. regulatory oversight of poultry processing since 2011a
Noticeb Date (mo-day-yr) Policy change Reference

76 FR 15282 03-21-2011 Salmonella and Campylobacter performance standards updated for establishments 23
processing young chickens and turkeys to accept 5 positive chicken samples per 51-
sample set as initially proposed by FSISc on 14 May 2010 (75 FR 27288).
78 FR 53017 08-28-2013 FSIS requested comment on adoption of a continuous sampling approach for 25
performance standard verification for all product classes.
79 FR 49566 08-21-2014 Poultry product regulations (9 CFR §381) (22) amended to establish a New Poultry 28
Inspection System (NPIS) for young chicken and turkey slaughter establishments. The
NPIS requires establishments to sort and remove unacceptable carcasses before
inspection by FSIS, shifts FSIS personnel from online inspection of carcasses to offline
inspection activities (e.g., verification of sanitation standard operating procedures),
requires establishments to demonstrate that product meets definition of ready-to-cook
poultry, and allows establishments to increase line speeds to 140 birds per min.
80 FR 3940 01-26-2015 Continuous sampling for performance standard verification adopted. Under this 29
approach, largest volume establishments are sampled approximately once per week.
Fewer samples are collected for smaller establishments. A “moving window”
approach for determining performance standard compliance adopted. Process control
categories (71 FR 9772) updated to reflect whether establishments meet a given
percentage of positive samples relative to the performance standard.
83 FR 56046 11-09-2018 Process control categories to be determined from the 52-week moving window ending 31
on the last Saturday of the previous month. The updated approach excludes follow-up
sampling results from process control category determination.
a
For major changes in regulatory oversight before 2011, see Ollinger et al. (13).
b
FR, Federal Register.
c
FSIS, USDA Food Safety and Inspection Service.

istics and Salmonella contamination (6, 11). However, these performance standard verification data from the Raw Chicken
studies were conducted before implementation of major Carcass Sampling Program obtained through the FSIS Web site;
changes to regulatory oversight of poultry processors (Table data on establishment characteristics obtained from the FSIS Meat,
1). In 2014, the FSIS began implementing the New Poultry Poultry and Egg Product Inspection (MPI) Directory; and FSIS
Inspection System (NPIS), which was designed to focus inspection and noncompliance data for individual establishments
obtained through a Freedom of Information Act request to FSIS.
FSIS efforts on inspection activities more directly related to
The primary outcome of interest was presence or absence of
public health outcomes such as process control systems for Salmonella in individual samples collected by FSIS inspectors.
Salmonella (28). Major changes to sampling protocols for Based on previous studies and preliminary analyses, the primary
performance standard verification have also been imple- risk factors considered were year, season, region (as defined by
mented. In previous risk factor studies, data from set-based FSIS Office of Field Operations District), whether the sample was
sampling (i.e., 51 samples collected in an establishment over collected in nBPW, annual slaughter volume, product class(es)
consecutive days) were used. However, in 2015 the FSIS slaughtered by establishment (e.g., turkey and other poultry),
(29) adopted a continuous sampling approach (i.e., approx- monthly processing volume, type of finished product (e.g., raw
imately one sample collected per establishment per week on intact and raw nonintact), and noncompliance with individual
a continuous basis). In 2016, the FSIS (30) also began regulations in Title 9 of the Code of Federal Regulations (21).
collecting carcass rinse samples with neutralizing buffered To address data sparsity issues, several variables were
dichotomized (Table 2). Slaughter volume was dichotomized to
peptone water (nBPW) to neutralize the antimicrobial
indicate whether an establishment slaughtered 10,000,000 birds
chemicals that are widely used in the poultry industry but
per year (i.e., the largest of five slaughter volume categories
negatively impact recovery of Salmonella from samples (7). assigned by FSIS within the MPI Directory). The FSIS Salmonella
Given these recent changes in sampling and regulatory verification testing program (25, 28, 29) includes a routine
oversight of poultry production, a reevaluation of risk factors sampling approach in which approximately five and two sample
for Salmonella contamination is warranted. The purpose of units (carcass rinses) are collected per month from higher volume
this study was to provide such an update by identifying risk establishments (i.e., 5,000,000 lb [2,267,962 kg] of product per
factors for Salmonella contamination of whole chicken month) and lower volume establishments (i.e., ,5,000,000 lb of
carcasses, such as establishment characteristics and inspec- product per month), respectively. To align with these sample
tion compliance history, through an evaluation of Salmonel- allocation practices, monthly processing volume (calculated by
la-positive test results among FSIS Salmonella performance multiplying mean daily processing volume by the number of
standard verification data. production days per month) was dichotomized to indicate a high or
low processing volume. Dichotomous variables were also created
to indicate hazard analysis critical control point (HACCP) product
MATERIALS AND METHODS
categories produced by each establishment.
Data source. Risk factors for Salmonella contamination of Establishment compliance was determined for 146 individual
whole chicken carcasses were evaluated based on Salmonella regulations by combining data on scheduled and performed
J. Food Prot., Vol. 84, No. 10 RISK FACTORS FOR SALMONELLA CONTAMINATION OF CHICKEN 1715

TABLE 2. Univariable association of establishment characteristics with Salmonella-positive results among whole chicken carcassesa
No. of No. of positive/ % positive Odds
Establishment characteristic establishmentsb total samples samples ratio 95% CI P (Wald)c

Geographic regiond
South West, District 05 7 81/1,326 6.11 1.948 1.184, 3.203 0.009*
North West, District 15 6 25/873 2.86 0.883 0.617, 1.262 0.493
West North Central, District 25 7 38/886 4.29 1.342 0.531, 3.389 0.534
MidWest Central, District 35 29 144/5,718 2.52 0.773 0.566, 1.057 0.107
South Central, District 40 22 226/4,350 5.20 1.641 1.008, 2.670 0.046*
East North Central, District 50 11 123/1,742 7.06 2.274 1.442, 3.587 ,0.001*
North Atlantic, District 60 14 136/1,978 6.88 2.210 1.423, 3.434 ,0.001*
MidAtlantic, District 80 37 329/7,520 4.38 1.370 0.980, 1.914 0.065
South Atlantic, District 85 29 303/6204 4.88 1.537 1.074, 2.200 0.019*
MidEast Central, District 90 41 320/9,900 3.23 REF
Slaughter volume (birds/yr)
10,000,000 157 1,395/37,313 3.74 0.336 0.261, 0.432 ,0.001*
,10,000,000 46 330/3,184 10.36 REF
Slaughter turkey
Yes 41 273/2,876 9.49 2.613 1.961, 3.481 ,0.001*
No 162 1,452/37,621 3.86 REF
Slaughter other poultry
Yes 26 194/1,972 9.84 2.636 1.967, 3.533 ,0.001*
No 177 1,531/38,525 3.97 REF
Processing volume (lb [kg] chicken/mo)
5,000,000 [2,267,961.85] 147 1,335/35,086 3.80 0.509 0.386, 0.671 ,0.001*
,5,000,000 56 390/5,411 7.21 REF
Produce raw nonintact chicken products
Ground, comminuted, mechanically separated
Yes 69 514/13,144 3.91 0.879 0.675, 1.143 0.334
No 134 1,211/27,353 4.43 REF
Injected, tenderized, vacuum tumbled
Yes 104 960/25,299 3.79 0.744 0.586, 0.945 0.015*
No 99 765/15,198 5.03 REF
Other
Yes 3 24/736 3.26 0.754 0.473, 1.202 0.235
No 200 1,701/39,761 4.28 REF
Produce raw intact chicken products
Whole Bird
Yes 126 1,121/26,454 4.24 0.985 0.778, 1.246 0.897
No 77 604/14,043 4.30 REF
Parts
Yes 187 1,653/38,727 4.27 1.051 0.618, 1.789 0.853
No 16 72/1,770 4.07 REF
Other
Yes 53 438/12,327 3.55 0.770 0.609, 0.972 0.028*
No 150 1,287/28,170 4.57 REF
Produce cooked, not shelf stable, RTE, poultry products
Yes 11 50/1,924 2.60 0.588 0.372, 0.929 0.023*
No 192 1,675/38,573 4.34 REF
Produce not fully cooked, not shelf stable, NRTE, poultry products
Yes 15 104/3,247 3.20 0.727 0.488, 1.084 0.118
No 188 1,621/37,250 4.35 REF
a
CI, confidence interval; REF, reference group; RTE, ready-to-eat; NRTE, not ready-to-eat.
b
Missing demographic factors led to exclusion of 22 establishments from the analytical data set. Eight establishments were no longer active;
three establishments changed from federal to state inspection during study period; eight establishments had addresses associated with
multiple names, identifications, or numbers; and three establishments were no longer listed in the MPI Directory with no known cause.
c
* P , 0.05.
d
As defined by the FSIS Office of Field Operations District.
1716 BECZKIEWICZ AND KOWALCYK J. Food Prot., Vol. 84, No. 10

inspection tasks with noncompliance records issued to establish-


ments. To align with the time frame used in determination of
public health regulations (PHRs) by the FSIS (32), a variable
indicating compliance, noncompliance, or nonobservance in the
84 days preceding sample collection was created for each
regulation (Supplemental Table S1).
Because seasonality has been associated with Salmonella
contamination, season was assigned for each observation using
week definitions from the Morbidity and Mortality Weekly Report:
winter, weeks 48 through 52 and 1 through 8 (roughly December
through February); spring, weeks 9 through 21 (roughly March
through May); summer, weeks 22 through 35 (roughly June
through August); and autumn, weeks 36 through 47 (roughly
September through November). To limit the impact of potential
uncontrolled confounders, establishments for which demographic
data were missing were excluded from analyses. A complete list
and description of variables in the analytical data set are provided
in Table S1. FIGURE 1. FSIS Salmonella performance standard verification
test results for whole chicken carcasses. Test results are reported
Statistical analyses and model development. Descriptive for 40,497 carcass rinse samples collected from 2015 through
statistics were estimated for establishment characteristics, micro- 2019. Proportion of samples testing positive for Salmonella is
biological results, and inspection data. Odds ratios (ORs) and 95% reported by year and season. This proportion should not be
confidence intervals (CIs) were estimated for potential risk factors interpreted as an industry-wide prevalence because verification
in the analyses. Appropriateness of the multivariable model was sampling programs were not designed for prevalence estimation.
evaluated using methods described by Preisser and Qaqish (15).
All analyses were conducted in SAS 9.4 (SAS Institute, Cary,
NC), and code for the multivariable risk factor model is provided RESULTS
in Supplemental Material S2.
A generalized estimating equation (GEE) approach was used Samples were collected from 225 establishments
to develop a multivariable risk factor model for Salmonella through the whole chicken carcass Salmonella verification
contamination of whole chicken carcasses (16, 37). Although the testing program from 2015 through 2019, but 22 establish-
GEE approach to statistical modeling requires no assumptions ments were excluded from this analysis because of missing
about the distribution of data beyond specifying an appropriate demographic factors. Of the remaining 203 establishments,
link (e.g., logit for binary data), a pairwise chi-square analysis of 72.4% slaughtered 10,000,000 birds per year (high
establishment demographic factors (e.g., slaughter volume and volume) and 27.6% slaughtered ,10,000,000 birds per
processing volume) was conducted to test for independence of year (low volume) (Table 2). A total of 40,497 carcass rinse
predictor variables and to inform the model building process. samples were collected and tested from the 203 establish-
Variables were tested for association with a Salmonella-positive
ments during the study period, and 1,725 (4.26%) were
test result using a working independence correlation structure.
positive for Salmonella, although this rate differed by year
Establishment characteristic variables were included in the
and season (Fig. 1). Of the 1,722 Salmonella isolates that
multivariable model selection process when marginally associated
(P , 0.20) with a Salmonella-positive result. Only 84-day were serotyped, 872 (50.6%) were Salmonella Kentucky,
historical compliance variables with ,10% of observations 244 (14.2%) were Salmonella Enteritidis, 168 (9.8%) were
classified as not observed and a marginally significant (P , Salmonella Infantis, 167 (9.7%) were Salmonella Typhi-
0.20) univariable association with a Salmonella-positive result murium, and 271 (15.7%) were identified as other
were considered in the multivariable process. serotypes. These proportions of samples positive for
The multivariable model was developed using a multistaged Salmonella should not be interpreted as an industry-wide
approach in a manner that minimized the quasi-likelihood under prevalence because verification sampling programs are not
the independence model criterion statistic first proposed by Pan designed for prevalence estimation. A total of 7,487,102
(14) for assessing GEE model fit (5). Ideally, all risk factors inspection tasks were scheduled among the 203 establish-
univariably associated with a Salmonella-positive result would be ments during the study period, resulting in a total of
jointly modeled with backward selection of the final multivariable 193,664 noncompliance records.
model. Nonconvergence of this model paired with a high number Eight of the 13 establishment demographic factors
of not observed values among historical compliance variables considered in the univariable GEE model were significantly
necessitated a modified approach. To avoid potential model
associated (P , 0.05) and 1 factor was marginally
misspecification from not observed inspection data, an initial
associated (P , 0.20) with a Salmonella-positive test result
multivariable model containing only establishment characteristics
and administrative information (e.g., year or season of sample (Table 2). Year and sample collection method (i.e., in nBPW
collection) was developed through backward selection of variables or not in nBPW) were significantly associated (P , 0.05)
marginally associated with a Salmonella-positive result. Historical and season was marginally associated (P , 0.20) with a
compliance variables that met the inclusion criteria were then Salmonella-positive result (Fig. 2). Of the 146 historical
sequentially added through forward selection and kept in the compliance variables evaluated by univariable analysis, 30
multivariable model when statistically significant (P , 0.05) were significantly associated (P , 0.05) and 25 were
according to the Wald test. marginally associated (P , 0.20) with a Salmonella-
J. Food Prot., Vol. 84, No. 10 RISK FACTORS FOR SALMONELLA CONTAMINATION OF CHICKEN 1717

avoid biasing other model estimates, this variable was


excluded from the final multivariable model. Of all factors
included in the model, the use of nBPW during sampling
was by far the strongest predictor of a Salmonella-positive
result.

DISCUSSION
As governmental regulatory agencies and food manu-
facturers work to continue improving the safety of meat and
poultry products, our understanding of how different
practices or factors impact the safety of such products also
needs improvement. In response to recent major changes in
U.S. regulatory oversight of poultry production, this study
was conducted to reevaluate risk factors for Salmonella
contamination of whole chicken carcasses. The results of
this study (i) confirmed that establishment volume and
FIGURE 2. Univariable association of sample-specific factors season remain risk factors for Salmonella contamination;
with Salmonella-positive test results among whole chicken carcass (ii) revealed new potential risk factors for Salmonella
samples. Odds ratios (with 95% confidence intervals) are reported. contamination; and (iii) revealed how accounting for
Each variable’s reference group is indicated by [ref]. Year (P ,
correlation between observations can impact results.
0.001) and sample collection in nBPW (P , 0.001) were
significantly associated with Salmonella-positive results. Season
(P ¼ 0.088) was marginally associated with Salmonella-positive Confirmation of prior risk factors. Following the
results. major changes to regulatory oversight of poultry, establish-
ment size remains a risk factor for Salmonella contamina-
positive result (Table S3). Only 22 of the 146 variables tion. Eblen et al. (6) found that large establishments (as
considered had a not observed value for ,10% of defined by the number of employees) had lower odds of
observations; 13 of these were at least marginally associated failing to meet the Salmonella performance standard than
with a Salmonella-positive result in the univariable analysis did small establishments. Our study, in which establishment
and were considered in the multivariable model (Table 3). size was defined by slaughter and processing volume,
Although exclusion of 124 regulation variables from further revealed that high-volume establishments had significantly
analyses was not ideal, doing so limits the potential for lower odds of a Salmonella-positive test result than did low-
model misspecification, which could negatively impact volume establishments. Although both studies found lower
generalizability of the results. Historical compliance odds of contamination among larger establishments, the
variables considered in the multivariable model pertained difference in how establishment size was defined in the two
to a wide range of regulations, including general labeling, studies is significant. HACCP-based size could be indica-
product weight labeling, poultry product inspection, sanita- tive of processing capacity (e.g., greater volume requires
tion, and HACCP. more employees to process), but defining establishment size
Backward selection of establishment demographic by either slaughter volume or processing volume provides a
factors followed by forward selection of historical compli- more accurate estimate of how much product is routing
ance variables resulted in a multivariable model with eight through an establishment. This approach would be espe-
factors significantly (P , 0.05) or marginally (P , 0.10) cially important for poultry processing in which scalding,
associated with a Salmonella-positive result (Table 4). As picking, and chilling of carcasses can lead to serial cross-
expected, odds of contamination were higher for samples contamination (3). Thus, the number of carcasses slaugh-
collected in summer than for those collected in winter. After tered and processed could truly impact contamination.
adjusting for seasonal and establishment-specific character- Because of data limitations, size was defined in the present
istics (e.g., annual slaughter volume), region was also study by dichotomous indicators for slaughter and process-
significantly associated with contamination. Higher odds of ing volume; these two variables were highly correlated
contamination were identified among establishments locat- (results not shown), and only one (slaughter volume) was
ed in the South West (District 05), South Atlantic (District kept in the model by the backward selection process.
85), and the East North Central (District 50) regions of the Regardless of how establishment size is defined, the
United States compared with establishments in the MidEast higher odds of contamination among establishments with
Central (District 90) region. Only two inspection variables lower slaughter volume is logical. For example, facilities
remained significant in the multivariable model: 9 CFR processing a larger volume are more likely to implement
§417.4 (HACCP system validation, verification, and processing technologies that improve safety (e.g., inside-
reassessment) and 9 CFR §317.2 (label definition and outside bird washer), microbiological testing, and food
required features) (21). Further investigation revealed that safety trainings for employees (4, 12). In contrast, lower
compliance with 9 CFR §317.2 provided minimal improve- volume facilities rely more on sanitation practices and
ment in model fit but led to exclusion of 2,265 observations modification to daily operations to maintain food safety
(5.6% of all observations) due to not observed values. To (12). Extending this work, an index for food safety
1718 BECZKIEWICZ AND KOWALCYK J. Food Prot., Vol. 84, No. 10

TABLE 3. Historical verification status and univariable association of 9CFR regulations (21) with Salmonella-positive test results for
whole chicken carcassesa
Historical verification status (84-day history)

Not compliant Compliant Not observed

9 CFR section No. % No. % No. % Odds ratio 95% CI P (Wald)b

317.2, Labeling, definition 841 2.0767 37,391 92.3303 2,265 5.5930 1.563 1.132, 2.160 0.007*
381.65, PP inspection, general 37,492 92.5797 3,005 7.4203 0 0 0.454 0.352, 0.585 ,0.001*
381.76, PP inspection, inspection 23,228 57.3573 16,831 41.5611 438 1.0816 0.881 0.724, 1.073 0.208
381.117, PP inspection, name of product 6,650 16.4210 32,934 81.3245 913 2.2545 1.081 0.858, 1.362 0.507
381.121, PP inspection, quantity 4,048 9.9958 34,720 85.7347 1,729 4.2695 1.327 0.898, 1.962 0.156
381.122, PP inspection, identification 1,747 4.3139 37,658 92.9896 1,092 2.6965 0.734 0.495, 1.087 0.122
381.123, PP inspection, official marks 2,856 7.0524 37,042 91.4685 599 1.4791 0.938 0.707, 1.244 0.657
381.125, PP inspection, handling 658 1.6248 38,001 93.8366 1,838 4.5386 1.087 0.604, 1.954 0.781
381.175, PP inspection, requirements 341 0.8420 38,099 94.0786 2,057 5.0794 1.861 1.047, 3.307 0.034*
416.1, Sanitation, rules 19,046 47.0306 21,450 52.9669 1 0.0025 0.978 0.814, 1.173 0.808
416.2, Sanitation, grounds and facilities 37,244 91.9673 3,084 7.6154 169 0.4173 0.691 0.521, 0.916 0.010*
416.4, Sanitation, sanitary operations 38,544 95.1774 1,628 4.0201 325 0.8025 0.55 0.403, 0.752 0.000*
416.5, Sanitation, employee hygiene 13,942 34.4272 24,690 60.9675 1,865 4.6053 1.05 0.878, 1.254 0.595
416.12, Sanitation, development SSOPs 4,277 10.5613 36,154 89.2757 66 0.1630 0.985 0.745, 1.304 0.919
416.13, Sanitation, implementation SSOPs 38,158 94.2243 2,339 5.7757 0 0 0.513 0.383, 0.687 ,0.001*
416.14, Sanitation, maintenance SSOPs 12,138 29.9726 28,326 69.9459 33 0.0815 1.098 0.905, 1.331 0.343
416.16, Sanitation, records 16,557 40.8845 23,940 59.1155 0 0 0.997 0.838, 1.185 0.972
417.2, HACCP systems, hazard analysis
and HACCP plan 28,204 69.6447 12,290 30.3479 3 0.0074 0.831 0.693, 0.998 0.047*
417.3, HACCP systems, corrective actions 9,159 22.6165 30,646 75.6747 692 1.7088 1.185 0.975, 1.440 0.089
417.4, HACCP systems, validation,
verification, reassessment 7,100 17.5322 33,395 82.4629 2 0.0049 1.333 1.104, 1.610 0.003*
417.5, HACCP systems, records 16,373 40.4302 24,122 59.5649 2 0.0049 1.101 0.953, 1.272 0.191
442.2, Weight labeling, net weight
compliance 4,190 10.3464 33,316 82.2678 2,991 7.3857 1.293 0.893, 1.872 0.173
a
Regulations presented when eligible for inclusion in multivariable model (,10% of observations not observed in 84-day compliance
history). Historical verification and univariable association with a positive Salmonella result for all regulations considered provided in
Table S3. CFR, Code of Federal Regulations; CI, confidence interval; PP, poultry product.
b
* P , 0.05.

technology among meat and poultry plants was developed, were found for 2016 and 2017 samples than for samples
and smaller volume facilities had lower technology indices from 2019. Although results from this study cannot be
than did larger volume facilities and facilities owned by used for causal interpretation, we hypothesized that the
multiplant firms (13). This higher technology index was apparent increase in Salmonella contamination following
also observed among meat and poultry plants that were introduction of nBPW encouraged establishments to
subject to food safety auditing (i.e., third-party food safety implement additional control measures that lead to
evaluations). contamination reductions over time. The FSIS continues
Season also remained a risk factor; the odds of to collect samples under the Salmonella verification testing
Salmonella contamination were higher in the summer than program, and follow-up evaluations of data collected after
in winter. Poultry contamination typically follows a 2018 could provide more insight into temporal trends and
seasonal pattern (35). Further temporality was seen with identify additional risk factors that may have been masked
decreasing odds of contamination across years, even after by the change to nBPW.
adjusting for other variables. The FSIS started using
nBPW for collecting samples in 2016 because of the Identification and reevaluation of risk factors.
negative impact of antimicrobial chemicals used in poultry Region and type of finished product were identified as
processing on Salmonella recovery from carcass rinses (7, new risk factors for Salmonella contamination. In contrast
36). Our results were consistent with these findings; the to previous studies, region (as defined by FSIS district) was
odds of testing positive for Salmonella were significantly significantly associated with a Salmonella-positive result in
higher for samples collected in nBPW. However, the the present study. The South West (District 05), South
introduction of nBPW did not account fully for the Atlantic (District 85), and East North Central (District 50)
temporal trends in contamination. After adjusting for regions had significantly higher odds of contamination than
nBPW use, significantly higher odds of contamination did the MidEast Central (District 90) region. The MidEast
J. Food Prot., Vol. 84, No. 10 RISK FACTORS FOR SALMONELLA CONTAMINATION OF CHICKEN 1719

TABLE 4. Multivariable risk factor models for association of establishment characteristics and 9 CFR regulations (21) with Salmonella-
positive test results for whole chicken carcassesa
Model variable Variable value Odds ratio 95% CI P (Wald)b

Year 2015 1.514 0.990, 2.315 0.056


2016 1.384 1.089, 1.758 0.008*
2017 1.252 1.010, 1.553 0.040*
2018 1.069 0.924, 1.238 0.368
2019 REF
Season Spring 1.043 0.899, 1.210 0.581
Summer 1.150 1.006, 1.314 0.040*
Fall 1.037 0.911, 1.180 0.584
Winter REF
Geographic regionc South West, District 05 1.541 1.043, 2.278 0.030*
North West, District 15 0.647 0.424, 0.990 0.045*
West North Central, District 25 0.980 0.513, 1.871 0.950
MidWest Central, District 35 0.807 0.604, 1.079 0.148
South Central, District 40 1.500 0.934, 2.409 0.094
East North Central, District 50 1.491 1.012, 2.198 0.043*
North Atlantic, District 60 1.321 0.804, 2.170 0.272
MidAtlantic, District 80 1.229 0.896, 1.687 0.201
South Atlantic, District 85 1.490 1.039, 2.137 0.030*
MidEast Central, District 90 REF
Sample collection nBPW 3.812 2.822, 5.149 ,0.001*
Non-nBPW REF
Slaughter volume 10,000,000 birds/yr 0.466 0.307, 0.710 ,0.001*
,10,000,000 birds/yr REF
Slaughter turkey Yes 1.557 0.973, 2.491 0.065
No REF
Produce cooked, not shelf stable, Yes 0.498 0.298, 0.833 0.008*
RTE poultry products No REF
Noncompliant with 9 CFR §417.4 Yes 1.249 1.071, 1.456 0.005*
(84-day history)d No REF
a
CFR, Code of Federal Regulations; CI, confidence interval; REF, reference group; nBPW, neutralizing buffered peptone water.
b
* P , 0.05.
c
As defined by the FSIS Office of Field Operations District.
d
Regulation applies to HACCP systems (validation, verification, and reassessment).

Central region was selected as the reference group because significantly correlated in this study (results not shown).
it has the highest concentration of poultry processing Further exploration of how the 11 establishments producing
establishments. These observed differences may have been raw and RTE finished products differed from other
due to differences in establishment characteristics or establishments could identify future areas for targeted
preharvest seasonal conditions, but because the differences improvement in Salmonella reduction.
remained after adjusting for seasonal conditions and One area that might help explain the inverse association
establishment characteristics, region may represent an between Salmonella contamination in raw product and
unmeasured risk factor. For example, region could be a regulation for RTE products is the establishment’s HACCP
surrogate measure for differences in management approach- system. In this study, the odds of a Salmonella-positive test
es between corporations and/or district offices. Given result among establishments noncompliant with any com-
sustained interest in food safety culture and the impact of ponent of HACCP system validation, verification, and
human behavior on food safety, collection of such data reassessment (9 CFR §417.4) (22) in the 84 days before
within regulatory contexts could facilitate inclusion of sample collection were higher than those for establishments
human factors and management approaches in future work that were compliant. Noncompliance with 9 CFR §417.4
on Salmonella contamination of meat and poultry products. could indicate a systematic process control issue for finished
Finished product type, which was not considered in products, and additional exploration of the relationship
previous studies, was associated with Salmonella contam- between HACCP noncompliance and Salmonella contam-
ination in the present study. Lower odds of contamination ination is warranted.
were found for establishments that produce both raw and None of the sanitation regulations univariably associat-
ready-to-eat (RTE) poultry products. Although larger ed with a Salmonella-positive result remained significant in
establishments typically process a wider variety of products, the final multivariable model, even though an association
slaughter volume and production of RTE products were not between sanitation noncompliance and Salmonella control
1720 BECZKIEWICZ AND KOWALCYK J. Food Prot., Vol. 84, No. 10

performance was found by Muth et al. (11). These authors the researchers addressed not observed values by including
defined compliance as the percentage of establishments only those historical variables with ,10% of observations
compliant for the specified inspection task code. In contrast, being not observed.
the present study used a categorical variable indicating 84- Although meat and poultry safety has improved since
day historical compliance, noncompliance, or not observed implementation of the pathogen reduction HACCP rule in
status for individual regulations. The approach used by Muth 1996 (22), Salmonella remains a major concern. The
et al. was, until recently, fairly common and used in risk proposed Healthy People 2030 goals continue to target a
assessments conducted by the FSIS (24, 27). However, our reduction in outbreaks of Salmonella infection (among other
approach aligns more closely with current methods used by foodborne pathogens) linked to poultry (33). Future policies
the FSIS to determine PHRs for meat and poultry processing and initiatives targeting the risk factors identified by our
on an annual basis (32) and, thus, identifies risk factors that GEE approach could reduce Salmonella contamination of
could be more easily targeted through policy initiatives. whole chicken carcasses and contribute to meeting the
Healthy People 2030 goals. Predictive modeling of
GEE approach to risk factor analyses. Regarding information collected on a regular basis, such as compliance
interpretation of our results in the context of previous work, with regulations as measured during daily inspection
odds ratios presented here are not directly comparable to activities, could inform directed real-time interventions for
those in previous risk factor studies conducted with U.S. controlling Salmonella contamination. Such an approach
regulatory data. Risk factor analyses conducted by Eblen et would likely be limited by not observed values among
al. (6) and Muth et al. (11) included a different outcome compliance data; therefore, future work in this area should
measure: a dichotomous variable indicating whether an explore alternative ways to model historical compliance
establishment met a given level of the Salmonella such as use of lagged and latent variables.
performance standard based on a set of 51 samples. This
difference is important because there are multiple ways to ACKNOWLEDGMENTS
address correlation between observations made in the same
The authors thank Isabel Walls, Eric Ebel, Rebecca Fields, Selena
establishment. In previous studies, a derived-variable Kremer, and Breauna Branch for providing input during the modeling
approach was used in which correlated observations (e.g., process and reviewing this manuscript. This research was supported by
51 samples collected within the same establishment) were The Ohio State University and an appointment to the FSIS Research
collapsed into a single observation and then analyzed using Participation Program administered by the Oak Ridge Institute for Science
independent data methods such as logistic regression. In and Education (ORISE) through an interagency agreement between the
U.S. Department of Energy (DOE) and the National Institute of
contrast, a GEE approach was used in the present study in
Environmental Health Sciences. ORISE is managed by Oak Ridge
which correlation between observations was addressed by Associated Universities (ORAU) under DOE contract DE-SC0014664.
postestimation adjustments to model parameter standard All opinions expressed in this article are the authors’ and do not
errors (9, 34, 37). This approach maintains a larger effective necessarily reflect the policies and views of USDA, FSIS, DOE, ORAU, or
sample size (e.g., 40,497 observations instead of the 203 ORISE.
observations obtained with the derived-variable approach),
which allows for a more granular analysis. Muth et al. (11) SUPPLEMENTAL MATERIAL
noted that analysis of regulatory data is often limited by the Supplemental material associated with this article can be
need for large data sets. As revealed in the present study, found online at: https://doi.org/10.4315/JFP-21-144.s1; https://doi.
this issue can be resolved by using a GEE approach to org/10.4315/JFP-21-144.s2
identify risk factors.
Several limitations to the present study were identified. REFERENCES
Twenty-two establishments were excluded from the analyt- 1. Alali, W. Q., and C. L. Hofacre. 2018. Preharvest food safety in
ical data set due to missing information for demographic broiler chicken production, p. 69–86. In S. Thakur and K. E. Kniel
variables. Missing data (e.g., no longer active or multiple (ed.), Preharvest food safety. American Society for Microbiology,
Washington, DC.
establishment names or identifications associated with an 2. Antunes, P., J. Mourao, J. Campos, and L. Peixe. 2016. Salmonel-
address) could have resulted from a lack of Salmonella losis: the role of poultry meat. Clin. Microbiol. Infect. 22:110–121.
control within the establishment. Although the primary goal 3. Byrd, J. A., and S. R. McKee. 2005. Improving slaughter and
of this study was to identify risk factors following processing technologies, p. 310–332. In G. C. Mead (ed.), Food
implementation of the NPIS, this study was not designed safety control in the poultry industry. CRC Press, Boca Raton, FL.
4. Cates, S. C., C. L. Viator, S. A. Karns, and M. K. Muth. 2007. Food
to assess the direct impact of the NPIS on Salmonella
safety practices of poultry slaughter plants: findings from a national
contamination within the poultry industry. The present study survey. Food Prot. Trends 27:954–966.
criteria for determining eligibility of historical inspection 5. Cui, J., and G. Qian. 2007. Selection of working correlation structure
variables were less restrictive than those used in other and best model in GEE analyses of longitudinal data. Commun. Stat.
studies. When determining PHRs, the FSIS requires that a Simul. Comput. 36:987–996.
regulation be verified at least 30 times within the historical 6. Eblen, D. R., K. E. Barlow, and A. L. Naugle. 2006. U.S. Food
Safety and Inspection Service testing for Salmonella in selected raw
window for inclusion in analyses; this stipulation is meant
meat and poultry products in the United States, 1998 through 2003:
to reduce the potential for biased values. Such a restriction an establishment-level analysis. J. Food Prot. 69:2600–2606.
was not applied in the present study because the vast 7. Gamble, G. R., M. E. Berrang, R. J. Buhr, A. Hinton, D. V. Bourassa,
majority of regulations would have been excluded. Instead, J. J. Johnston, K. D. Ingram, E. S. Adams, and P. W. Feldner. 2016.
J. Food Prot., Vol. 84, No. 10 RISK FACTORS FOR SALMONELLA CONTAMINATION OF CHICKEN 1721

Effect of simulated sanitizer carryover on recovery of Salmonella comments and announcement of implementation schedule. Fed.
from broiler carcass rinsates. J. Food Prot. 79:710–714. Regist. 76:15282–15290.
8. Heyndrickx, M., D. Vandekerchove, L. Herman, I. Rollier, K. 24. U.S. Department of Agriculture, Food Safety and Inspection Service.
Grijspeerdt, and L. De Zutter. 2002. Routes for Salmonella 2012. FSIS risk assessment for guiding public health-based poultry
contamination of poultry meat: epidemiological study from hatchery slaughter inspection. U.S. Department of Agriculture, Food Safety
to slaughterhouse. Epidemiol. Infect. 129:253–265. and Inspection Service, Washington, DC.
9. Huber, P. J. 1967. The behavior of maximum likelihood estimates 25. U.S. Department of Agriculture, Food Safety and Inspection Service.
under nonstandard conditions, p. 221–233. In L. Lecam and J. 2013. Changes to the Salmonella verification sampling program:
Neyman (ed.), Proceedings of the Fifth Berkeley Symposium on analysis of raw beef for Shiga toxin–producing Escherichia coli and
Mathematical Statistics and Probability. University of California Salmonella. Fed. Regist. 78:53017–53020.
Press, Berkeley. 26. U.S. Department of Agriculture, Food Safety and Inspection Service.
10. Mainali, C., G. Gensler, M. McFall, R. King, R. Irwin, and A. 2014. California firm recalls chicken products due to possible
Senthilselvan. 2009. Evaluation of associations between feed Salmonella Heidelberg contamination. Available at: https://www.fsis.
withdrawal and other management factors with Salmonella contam- usda.gov/wps/portal/fsis/topics/recalls-and-public-health-alerts/
ination of broiler chickens at slaughter in Alberta. J. Food Prot. recall-case-archive/archive/2014/recall-044-2014-release. Accessed
72:2202–2207. 8 January 2021.
11. Muth, M. K., M. Fahimi, and S. A. Karns. 2009. Analysis of 27. U.S. Department of Agriculture, Food Safety and Inspection Service.
Salmonella control performance in U.S. young chicken slaughter and 2014. FSIS risk assessment for guiding public health-based poultry
pork slaughter establishments. J. Food Prot. 72:6–13. slaughter inspection: an implementation scenario-based approach.
12. Ollinger, M., D. Moore, and R. Chandran. 2004. Meat and poultry U.S. Department of Agriculture, Food Safety and Inspection Service,
plants’ food safety investments: survey findings. TB-1911. U.S. Washington, DC.
Department of Agriculture, Economic Research Service, Washing- 28. U.S. Department of Agriculture, Food Safety and Inspection Service.
ton, DC. 2014. Modernization of poultry slaughter inspection. Fed. Regist.
13. Ollinger, M. E., M. K. Muth, S. A. Karns, and Z. Choice. 2011. Food 79:49566–49637.
safety audits, plant characteristics, and food safety technology use in 29. U.S. Department of Agriculture, Food Safety and Inspection Service.
meat and poultry plants. EIB-82. U.S. Department of Agriculture, 2015. Changes to the Salmonella and Campylobacter verification
Economic Research Service, Washington, DC. testing program: proposed performance standards for Salmonella and
14. Pan, W. 2001. Akaike’s information criterion in generalized Campylobacter in not-ready-to-eat comminuted chicken and turkey
estimating equations. Biometrics 57:120–125. products and raw chicken parts and related agency verification
15. Preisser, J. S., and B. F. Qaqish. 1996. Deletion diagnostics for procedures and other changes to agency sampling. Fed. Regist.
generalised estimating equations. Biometrika 83:551–562. 80:3940–3950.
16. Prentice, R. L. 1988. Correlated binary regression with covariates 30. U.S. Department of Agriculture, Food Safety and Inspection Service.
specific to each binary observation. Biometrics 44:1033–1048. 2016. New neutralizing buffered peptone water to replace current
17. Rayes, D. L. 2018. Craten v. Foster Poultry Farms Inc. CV-15-02587- buffered peptone water for poultry verification sampling. FSIS notice
PHX-DLR (District of Arizona, 19 January 2018). 41-16. U.S. Department of Agriculture, Food Safety and Inspection
18. Scallan, E., R. M. Hoekstra, F. J. Angulo, R. V. Tauxe, M.-A. Service, Washington, DC.
Widdowson, S. L. Roy, J. L. Jones, and P. M. Griffin. 2011. 31. U.S. Department of Agriculture, Food Safety and Inspection Service.
Foodborne illness acquired in the United States—major pathogens. 2018. Changes to the Salmonella and Campylobacter verification
Emerg. Infect. Dis. 17:7–15. testing program: revised categorization and follow-up sampling
19. Scharff, R. L. 2020. Food attribution and economic cost estimates for procedures. Fed. Regist. 83:56046–56049.
meat- and poultry-related illnesses. J. Food Prot. 83:959–967. 32. U.S. Department of Agriculture, Food Safety and Inspection Service.
20. Tack, D. M., L. Ray, P. M. Griffin, P. R. Cieslak, J. Dunn, T. Rissman, 2020. FY2021 public health regulations. U.S. Department of
R. Jervis, S. Lathrop, A. Muse, M. Duwell, K. Smith, M. Tobin- Agriculture, Food Safety and Inspection Service, Washington, DC.
D’Angelo, D. J. Vugia, J. Zablotsky Kufel, B. J. Wolpert, R. Tauxe, 33. U.S. Department of Health and Human Services. 2021. Healthy
and D. C. Payne. 2020. Preliminary incidence and trends of People 2030. Available at: https://health.gov/healthypeople. Ac-
infections with pathogens transmitted commonly through food— cessed 18 March 2021.
Foodborne Diseases Active Surveillance Network, 10 U.S. sites, 34. White, H. 1980. A heteroskedasticity-consistent covariance matrix
2016–2019. Morb. Mortal. Wkly. Rep. 69:509–514. estimator and a direct test for heteroskedasticity. Econometrica
21. U.S. Department of Agriculture, Food Safety and Inspection Service. 48:817–838.
n.d. Animals and animal products. 9 CFR chap. III, parts 300–599. 35. Williams, M. S., E. D. Ebel, N. J. Golden, and W. D. Schlosser. 2014.
U.S. Department of Agriculture, Food Safety and Inspection Service, Temporal patterns in the occurrence of Salmonella in raw meat and
Washington, DC. poultry products and their relationship to human illnesses in the
22. U.S. Department of Agriculture, Food Safety and Inspection Service. United States. Food Control 35:267–273.
1996. Pathogen reduction; hazard analysis and critical control point 36. Williams, M. S., E. D. Ebel, S. A. Hretz, and N. J. Golden. 2018.
(HACCP) systems; final rule. 9 CFR 304, 308, 310, 320, 327, 381, Adoption of neutralizing buffered peptone water coincides with
416, and 417. Fed. Regist. 61:38806–38989. changes in apparent prevalence of Salmonella and Campylobacter of
23. U.S. Department of Agriculture, Food Safety and Inspection Service. broiler rinse samples. J. Food Prot. 81:1851–1863.
2011. New performance standards for Salmonella and Campylobac- 37. Zeger, S. L., and K.-Y. Liang. 1986. Longitudinal data analysis for
ter in young chicken and turkey slaughter establishments: response to discrete and continuous outcomes. Biometrics 42:121–130.

You might also like