Preventive Veterinary Medicine 114 (2014) 73–87
Contents lists available at ScienceDirect
Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed
Description of the pig production systems, biosecurity practices and herd health providers in two provinces with high swine density in the Philippines J.I. Alawneh a,∗ , T.S. Barnes b , C. Parke a , E. Lapuz c , E. David c , V. Basinang d , A. Baluyut e , E. Villar f , E.L. Lopez g , P.J. Blackall a a
School of Veterinary Science, University of Queensland, Gatton 4343, Queensland, Australia Queensland Alliance for Agriculture and Food Innovation, University of Queensland, St Lucia 4067, Queensland, Australia c Department of Agriculture Region 3, RADDL, Pampanga, Philippines d Provincial Veterinary Office of Bulacan, Malolos City, Bulacan, Philippines e Provincial Veterinary Office of Pampanga, City of San Fernando, Pampanga, Philippines f Livestock Research Division PCAARRD-DOST, Los Ba˜ nos, Laguna, Philippines g Animal Health Division, Bureau of Animal Industry, Quezon City, Philippines b
a r t i c l e
i n f o
Article history: Received 23 July 2013 Received in revised form 16 January 2014 Accepted 20 January 2014
Keywords: Pig Swine Biosecurity Social network analysis Herd health Risk-based disease control
a b s t r a c t A cross-sectional study was conducted between October 2011 and March 2012 in two major pig producing provinces in the Philippines. Four hundred and seventy one pig farms slaughtering finisher pigs at government operated abattoirs participated in this study. The objectives of this study were to group: (a) smallholder (S) and commercial (C) production systems into patterns according to their herd health providers (HHPs), and obtain descriptive information about the grouped S and C production systems; and (b) identify key HHPs within each production system using social network analysis. On-farm veterinarians, private consultants, pharmaceutical company representatives, government veterinarians, livestock and agricultural technicians, and agricultural supply stores were found to be actively interacting with pig farmers. Four clusters were identified based on production system and their choice of HHPs. Differences in management and biosecurity practices were found between S and C clusters. Private HHPs provided a service to larger C and some larger S farms, and have little or no interaction with the other HHPs. Government HHPs provided herd health service mainly to S farms and small C farms. Agricultural supply stores were identified as a dominant solitary HHP and provided herd health services to the majority of farmers. Increased knowledge of the routine management and biosecurity practices of S and C farmers and the key HHPs that are likely to be associated with those practices would be of value as this information could be used to inform a risk-based approach to disease surveillance and control. © 2014 Elsevier B.V. All rights reserved.
1. Introduction The Philippine swine population was estimated at around 1.6 million (M) sows with a total of 12.2 M pigs
∗ Corresponding author. Tel.: +61 467 601055; fax: +61 7 54601992. E-mail addresses: [email protected], ibr [email protected], [email protected] (J.I. Alawneh). 0167-5877/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.prevetmed.2014.01.020
(BAS, 2011). The majority of the pig population is on the islands of Luzon (47%; 5.7 M) and Mindanao (29%; 3.5 M). A key feature of the Philippine pig industry is the smallholder production system (termed smallholder farms). Smallholder farmers (S), as defined by the Philippines Bureau of Agricultural Statistics (BAS) and used for the purpose of this study, were farmers raising one to 41 pigs (young or adult) in their backyards per year (BAS, 2011). Approximately 70% (8.5 M) of the total pig population is found on S
74
J.I. Alawneh et al. / Preventive Veterinary Medicine 114 (2014) 73–87
farms (BAS, 2011), and pigs raised are often used for their own consumption or as financial security (More et al., 1999; Lanada et al., 2005; Lee et al., 2005). Commercial farms (C) as defined by BAS and used for the purpose of this study are farms which satisfy at least one of the following conditions (BAS, 2011): (1) at least 21 head of adult pigs; (2) at least 41 head of grower pigs; or (3) at least 10 head of adults and 22 head of grower pigs. The commercial production system represents 30% of the Philippine pig industry. Smallholder and C farms almost exclusively supply pork to the local fresh meat markets indirectly through livestock traders (Bantugan et al., 1992; More et al., 1999). On a daily basis, livestock traders source their pigs from multiple farms. The daily activity of livestock traders is mainly driven by fresh meat market demands. The structure of the pig industry in the Philippines is likely to increase the frequency of contact (direct and indirect) between herds (Hurnik et al., 1994; Stark, 2000; Ribbens et al., 2009). This increased frequency of contact increases the likelihood and severity of introducing respiratory disease agents irrespective of production type (Gardner et al., 2002; Otake et al., 2010; Lambert et al., 2012b). The consequences of disease introduction were manifested in the porcine reproductive and respiratory syndrome (PRRS) and porcine circovirus type 2 (PCV2) outbreaks in 2008 (Resontoc, 2009). Both S and C farms reported increased mortality in their herds and this resulted in substantial financial and domestic trade implications for the Philippine swine industry (Resontoc, 2009; BAS, 2011). With the silent circulation of infectious agents like PCV2 and PRRS virus, the challenge for regulatory veterinary authorities, farm management and their herd health providers (HHP) is to ensure that sufficient measures (e.g. biosecurity, housing, vaccination, transport, market access) are always in place to reduce the risk of disease introduction into herds (Amass and Clark, 1999; Gardner et al., 2002; Lambert et al., 2012b). There is a paucity of information about farm management, animal husbandry and biosecurity practices of the S and C production systems in the Philippines. Knowledge of ‘weak points’ or areas of ‘vulnerability’ of production systems (‘vulnerability’ can be a subjective assessment of certain management practices such as biosecurity) where infectious pathogens might enter the herd and establish is important because it provides a better focus for disease control and surveillance activities. Filipino pig farmers have access to several HHPs that range in training and experience from agricultural supply store personnel to private swine veterinary consultants. It is plausible to assume that the effectiveness of farm management practices to reduce the risk of disease introduction (or maintenance) in the herd is influenced by farm choice of HHP. Therefore, in a developing country such as the Philippines where financial resources are limited, a better understanding of farm and HHP network dynamics has the potential to form the basis of a risk-based surveillance or disease control programme. Local veterinary authorities could identify ‘vulnerable’ farms and their HHP(s) through routine field investigations. The authorities could then preferentially communicate with HHP(s) of ‘vulnerable’ farms to improve the management practices of the
identified HHPs’ entire client network rather than focusing on individual farms. Social network analysis (SNA; Wasserman and Faust, 1994) offers a means for formalising the process of identifying ‘vulnerable’ farms and their HHPs. Social network analysis provides a network-based approach to reveal the structure of the relationships between farm and HHP (Dube et al., 2009), and to quantify those relationships within the network (Wasserman and Faust, 1994). The use of SNA in animal populations is increasing (Christley and French, 2003; Christley et al., 2005; Martin et al., 2011), and reviews are provided by Newman (2003) and Dube et al. (2009). The objectives of this study were to: (a) group S and C production systems into patterns according to their HHP(s); (b) obtain descriptive information about the grouped S and C production systems in two provinces of high swine density in the Philippines; and (c) identify key HHPs within each production system using SNA. Increased knowledge of the characteristics of S and C farms (e.g. farm type, size, and biosecurity practices) and the key HHPs that are likely to be associated with or influence those characteristics would be of value as this information could be used to inform a risk-based approach to disease surveillance and control. 2. Materials and methods 2.1. Study design and source population A cross-sectional study was conducted between October 2011 and March 2012 in two major pig producing provinces (Bulacan and Pampanga) in Region 3 (Central Luzon). Region 3 was selected because it is one of the most intensive pig rearing regions in the Philippines (BAS, 2011), and has reported major outbreaks and losses from acute respiratory disease in pigs in 2008 PRRS and PCV2 outbreaks (Resontoc, 2009). This study is part of a large abattoir based cross-sectional study that aimed to estimate prevalence and risk factors associated with respiratory disease lesions in finisher pigs at slaughter. Therefore, the target population was pig farms producing finisher pigs in Bulacan and Pampanga provinces. The source population was farms slaughtering finisher pigs on randomly selected calendar dates at government operated abattoirs (n = 29) in Bulacan and Pampanga provinces. Because a complete list comprising source and number of slaughtered finisher pigs was not available prior to slaughter time, systematic random sampling was used at the time of slaughter to select the study population (farms of origin). 2.2. Farm enrolment process Field staff of the Provincial Veterinary Office (PVO) asked meat inspectors at participating slaughterhouses and livestock traders in Bulacan and Pampanga provinces to record the following information at the time that finisher pigs arrived for slaughter: (1) date of slaughter; (2) livestock trader name, address and contact details; (3) livestock transporter name, address and contact details; (4) name and address of the pig farm from which pigs were purchased; (5) date of purchase; and (6) number of pigs
J.I. Alawneh et al. / Preventive Veterinary Medicine 114 (2014) 73–87
purchased. Information provided by the livestock traders to PVO field staff was verified independently by a trained surveyor team (see questionnaire section) and by the named pig farms within two weeks from slaughter date. Farm participation was on a voluntary basis. Because the large study aimed to estimate prevalence and risk factors associated with respiratory disease lesions at slaughter, pig farms were excluded (28%, 184/655; Fig. 1) if there was inaccuracy in the livestock trader’s supplied pig farm details (mainly farm location especially if those farms were located at the borders between Bulacan and Pampanga; 8%, 54/655), pig farms were located in other provinces (4%, 26/655), farms were untraceable (12%, 75/655), or refused to participate (4%, 29/655). Moreover, a farm would also have been excluded (0%, 0/655) if the date of slaughter occurred more than six weeks after the pigs were sold to a livestock trader. A maximum six-week interval was selected based on the fact that respiratory lesions (particularly Mycoplasma hyopneumoniae infection) in pig lungs can resolve eight to 12 weeks after they develop (Kobisch et al., 1993; Fano et al., 2005). 2.3. Farm questionnaire and data management A questionnaire was designed that consisted of three main sections. The first section collected general information about the interviewee (farm owner/manager or person directly working with the pigs on-farm; termed farmer) and farm profiles, such as the farmer demographics and level of education, farm location, type of production system (S or C), farm HHP (semi-open ended question; Table 1), animal husbandry and reproduction management. The second section requested information on the following: whether pigs on the farm come in contact with wild animals or pigs from neighbouring farms, number of workers living on the farm, swill feeding and ‘feedback’ practice (feeding pig manure, viscera or aborted material to pigs to increase herd immunity). Farmers were asked about various farm biosecurity measures: pen hygiene and cleaning frequency, cleaning method and cleaning products used, disposal method for removal of animal waste and faeces, whether traders and their vehicles were allowed to enter the farm and management practices put in place to prevent introduction of disease by vehicles and visitors, ability of farm workers to recognise sick and healthy pigs, animal health monitoring and management, and the interviewees awareness of pig density within a 500 metre radius relative to the farm. The third section asked for information regarding pig management comprising: type of pig housing, source of water supply for the farm and herd size details at the time of the interview, details of pig feed, vaccination routine, and history of any respiratory disease outbreak or outbreak investigation carried out on the farm in the last 12 months. In total, the questionnaire consisted of 71 questions, of which 10 (14.0%) were open, 13 (18.3%) were semi-open and 48 (67.7%) were closed questions with mostly dichotomous answers. Questions (n = 6) with distinct subordinate parts had at least five possible options. The responses were collected through face-to-face interviews conducted by four experienced interviewers
75
(two teams each comprising two interviewers) between October 2011 and March 2012. As there are different dialects in the Philippines, the questionnaire was written in English and translated to the appropriate dialect at the interview. To reduce information bias the questionnaire was pretested on experts in the Philippines pig production systems comprising regional and provincial veterinary officers and animal health advisors. All questions in the questionnaire were clarified with all interviewers before the study start date. The interviewers were instructed to ask questions exactly as stated in the questionnaire and provide only non-directive guidance. To minimise inter-observer variability in conducting the interview, all observers and PVO personnel met after the questionnaire was piloted on the six farms to agree on a common interpretation of the findings. If there was disagreement, the interpretation of the PVO personnel was chosen. To minimise information (misclassification) and selection biases, the interviewees were asked to verify the trader’s identity, date when the pigs were sold and number of pigs sold for slaughter before an interview was conducted (Fig. 1). The validity of the collected questionnaire data was confirmed during follow-up visits to six farms (three in each province) by the first author, the interviewers and Provincial Veterinary Officers personnel. To reduce misclassification bias that could arise from coding errors, the interviewers and the first author checked and corrected impossible coding of categorical variables (n = 80) and unreliable outlier values for continuous variables (n = 3). Throughout the study period questionnaire data were transferred three times a week (Monday, Wednesday, and Friday) to a purpose built relational database (MicrosoftTM Access® 2010, Microsoft Corporation). 2.4. Statistical analysis The unit of analysis was the individual farm. Continuous variables are described using mean, median, minimum, maximum values, 1st and 3rd quartiles, while categorical variables are presented as percentages. Farm profile, farmer demographics, and farm management practice data were described, and the associations with production type were assessed using Fisher’s exact 2 test for categorical variables or Wilcoxon rank sum test to test the equality of the medians for continuous variables. The tests were conducted using R (R Development Core Team, 2012). 2.4.1. Grouping farms based on herd health providers Agglomerative hierarchical cluster analysis (Ward, 1963; Everitt et al., 2001) was used to identify similar subgroups of farms (S and C) ‘homogenously’ clustered based on their HHP(s). Using an iterative procedure, at each stage squared distances between clusters were recomputed by the Lance–Williams dissimilarity update formula according to Ward’s minimum variance method. Ward’s method was selected because clusters are merged based on a minimum increase in total within-cluster variance after merging, therefore resulting in compact and spherical clusters (Ward, 1963; Everitt et al., 2001). To assess the certainty in the clustering, a thousand bootstrap replications were run for each cluster and a probability value
76
J.I. Alawneh et al. / Preventive Veterinary Medicine 114 (2014) 73–87
Fig. 1. Flowchart of farm (smallholder[S] and commercial[C]) enrolment, numbers and reasons for exclusion, number of farms eligible, and farms available for the final analysis through the study period between October 2011 and March 2012in two provinces in the Philippines.
(approximately unbiased (AU) P-value) was generated through multiscale bootstrap resampling. The agglomerative coefficient (AC) was also calculated to measure the clustering structure found by this technique. Both the AU P-value and AC values can range from 0 to 1 with higher values indicative of greater confidence in the clustering structure. The significance of the clusters was set at 0.9 for the AU P-value of the clusters. The AU P-value and AC values were derived using the Pvclust (Suzuki and Shimodaira, 2006) and Cluster (Maechler et al., 2012) packages in R. 2.4.2. Farm and herd health providers networks Five matrices were constructed. In these matrices farmers formed rows (in SNA terminology farmers would be called ‘actors’ and HHPs would be called ‘affiliates’). The interaction between farmers and their HHPs formed the ‘ties’ in the matrices. Ties were considered as undirected and were dichotomised, 1 indicating the presence of the tie and 0 otherwise. Farmers without a HHP (i.e. isolated; n = 3) were identified and removed before further analyses were conducted. Social network metrics, analyses and figures were derived using R, Ucinet 6.421 (Borgatti et al., 2002), and NetDraw 2.123 (Borgatti et al., 2002). A focus of the current study was to explore the network of the farmers and then use that network to identify the main HHP(s) (‘more centrally located’) in the overall network and within each of the identified farmer group network structures. Thus the eigenvector and normalised
eigenvector centrality measures for a 2-mode network were calculated. Eigenvector centrality is a weighted degree measure in which the centrality of a farm/HHP is proportional to the sum of centralities of the HHP/farm it is connected to. The normalised eigenvector centrality is the scaled eigenvector centrality divided by the maximum difference possible in a given network expressed as a percentage (Borgatti et al., 2002). The eigenvector value is a measure of farmer or HHP importance based on their activity, position, and contacts or popularity in the network. Because each farmer could have more than one HHP, it was of interest to determine whether pairs of farmers who engage with one HHP are more likely to also engage with another. To do this, 10 matrices each were constructed for the S and C production systems. Each of these symmetric matrices was specific to one of the 10 HHPs and consisted of rows and columns equal to the number of farmers in the production system. A connection between each farmer pair (of the same production type) was coded as 1. Jaccard’s coefficients were then estimated for HHP matrix pairs using the Quadratic Assignment Procedure (QAP; Hubert and Arabie, 1989) with a permutation-based significance test involving 5000 random permutations. Jaccard’s coefficient, measures the ‘similarity’ of farmer ties (binary) between HHP matrix pairs and is expressed as a proportion where a value of 1 indicates complete similarity and 0 indicates complete dissimilarity. The QAP randomly permutes rows and columns (synchronously) of one matrix
J.I. Alawneh et al. / Preventive Veterinary Medicine 114 (2014) 73–87
77
Table 1 A complete list of herd health providers (n = 10; herd health providers abbreviation used in throughout the manuscript in parentheses) included in the farm questionnaire conducted in two provinces in the Philippines between October 2011 and March 2012. The list was presented to the interviewees (n = 471) as a semi-open ended question with multiple choices allowed. Herd health provider
Definition
Technical consultant contract grower (TCCG)
A veterinarian providing herd health services related to pig health and performance in the growing stage A veterinarian from the government or the private sector contracted to provide herd health service and monitor farm performance at all stages of production cycle A technical-level government employee who implements local and national livestock programmes. These individuals conduct routine rounds in their areas of responsibility to inspect and monitor the health of livestock and extend veterinary services to smallholder farms A veterinarian who resides on the farm as farm manager and provides herd health service at all stages of production cycle An organised group of pig farmers who operate autonomously and privately as a national union or a local cooperative to support member interests in swine production. It has concrete programmes in swine production including credit and marketing assistance for members A government-employed veterinarian who conducts animal disease surveillance (passive surveillance) and investigates reported disease outbreaks in the field Neighbour pig farmer (usually among smallholders), a co-member of the Hog farmer association, or a co-member of the Farmer co-operative An organised group of pig farmers who operate autonomously and privately as a national union or a local cooperative to support member interests in swine production. It has concrete programmes in swine production including credit and marketing assistance for members An individual with technical-level expertise who work for multi-national pharmaceutical companies to promote and sell their products. They provide technical information about their products as well as technical services such as diagnostic services relative to the purchases made by farms with large operations and feed mills A retailer – with or without veterinary or technical qualifications – of agricultural and veterinary supplies such as pharmaceutical drugs, supplements, feed and mixes for livestock who operate in towns and is accessible to farmers. Retailer activities are not monitored nor do they report to any regulatory veterinary authority Other herd health provider not listeda
Private veterinarian consultant (PC) Livestock inspector–agricultural technician (LIAT)
On-farm veterinarian (OFV) Hog farmer association (HFA)
Government veterinarian (GV) Fellow pig farmer (FPF) Farmer co-operative (FC)
Pharmaceutical company representative (PCR)
Agricultural supply store (AS)
Other a
No other herd health providers were listed by farmers.
and recomputes Jaccard’s coefficients in order to compute the proportion of times (per n permutations) that a random measure is larger than or equal to the observed. A low proportion ( 0.90) within each production system (S farm clusters A and B, termed SA and SB respectively; C farm clusters A and B, termed CA
and CB respectively). Descriptive statistics of farm profile, farmer profile, and details of farmers’ choice of HHP(s) according to production type are shown in Table 2. Overall, S and C farmers had different (P < 0.01) combinations of HHP. Smallholder farmers tended to use agriculture supply store (AS; 63%), Livestock inspector–agricultural technician (LIAT; 50%), and government veterinarian (GV; 28%) as their preferred HHP(s), while commercial farms were more reliant on private veterinary services; 61% used private consultants (PC), 51% used on farm veterinarian (OFV) and 48% used pharmaceutical company representatives (PCR). Within S farms, the distribution of farms into the two clusters differed (P < 0.01) between provinces. Compared with SB (n = 275), SA (n = 96) had on average more productive sows (P < 0.01), more pigs on the farm (P < 0.01), were more likely to house pigs in separate pens (P < 0.01) rather than free range (P < 0.01), used a private veterinarian (P < 0.01) as the main HHP, and were less dependent on pig farming as a main source of income (P < 0.01). For C farms, CB (n = 33) had fewer pigs (P < 0.01) and were more likely to use livestock inspectors (P < 0.01), pharmaceutical company representatives (P = 0.02), government veterinarians (P = 0.01), and AS (P < 0.01) as their HHP compared with CA (n = 67). Descriptive statistics of farm management and biosecurity practices are presented in Tables 3 and 4. Compared with S farms, overall, C farms showed higher frequency
78
J.I. Alawneh et al. / Preventive Veterinary Medicine 114 (2014) 73–87
Table 2 Descriptive statistics of farm (n = 471) and farm manager profiles classified by production type (smallholder, n = 371; commercial, n = 100) and herd health provider (smallholder A, SA, n = 96; smallholder B, SB, n = 275; commercial A, CA, n = 67; and commercial B, CB, n = 33) for participating pig farms from two provinces in Philippines between October 2011 and March 2012. Profile
Smallholder farms SA n (%)
Farm profile Province Bulacan Pampanga Number of productive sows Median; Min, Max Mean; 1st quartile, 3rd quartile Total number of animals Median; Min, Max Mean; 1st quartile, 3rd quartile Housing type Separate pens Pen adjacent to house Free range Other Herd health provider None Technical consultant contract grower Private veterinarian (consultant) Livestock inspector–agricultural technician On-farm veterinarian Hog farmer association Government veterinarian Fellow pig farmer Farmer co-operative Pharmaceutical company representative Agricultural supply store Farm manager profile Gender Female Male Education Elementary or cannot read High school College Years of experience Median (Min, Max) Mean (1st quartile, 3rd quartile) Proportion of income derived from raising pigs 75% Unknown
SB n (%)
52 (54) 44 (46)
113 (40) 162 (60)
0; 0, 30 3; 0, 3
0; 0, 18 1; 0, 0
14; 1, 92 22; 4, 11
6; 0, 77 9; 4, 11
61 (64) 32 (33) 1 (1) 2 (2)
120 (44) 87 (32) 61 (22) 7 (3)
2 (2) 4 (4) 72 (75) 40 (42) 4 (4) 1 (1) 1 (1) 1 (1) 1 (1) 6 (6) 14 (15)
1 (0) 3 (1) 0 (0) 145 (53) 4 (2) 0 (0) 103 (38) 61 (22) 3 (1) 8 (3) 220 (80)
28 (29) 68 (71)
84 (31) 191 (69)
11 (12) 45 (47) 40 (42)
25 (9) 154 (56) 96 (35)
7 (1, 39) 10 (5, 12)
7 (1, 42) 9 (5, 9)
39 (41) 19 (20) 8 (8) 2 (2) 28 (29)
79 (29) 95 (35) 2 (1) 1 (0) 98 (36)
Commercial farms a
P
0.03
CA n (%)
S vs. C CB n (%)
24 (36) 43 (64)
18 (55) 15 (45)
223; 0, 2580 395; 121, 425
1; 0, 350 39; 0, 35
1520; 1, 19,630 3102; 629, 3232
22; 0, 6439 481; 6, 382
Contents lists available at ScienceDirect
Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed
Description of the pig production systems, biosecurity practices and herd health providers in two provinces with high swine density in the Philippines J.I. Alawneh a,∗ , T.S. Barnes b , C. Parke a , E. Lapuz c , E. David c , V. Basinang d , A. Baluyut e , E. Villar f , E.L. Lopez g , P.J. Blackall a a
School of Veterinary Science, University of Queensland, Gatton 4343, Queensland, Australia Queensland Alliance for Agriculture and Food Innovation, University of Queensland, St Lucia 4067, Queensland, Australia c Department of Agriculture Region 3, RADDL, Pampanga, Philippines d Provincial Veterinary Office of Bulacan, Malolos City, Bulacan, Philippines e Provincial Veterinary Office of Pampanga, City of San Fernando, Pampanga, Philippines f Livestock Research Division PCAARRD-DOST, Los Ba˜ nos, Laguna, Philippines g Animal Health Division, Bureau of Animal Industry, Quezon City, Philippines b
a r t i c l e
i n f o
Article history: Received 23 July 2013 Received in revised form 16 January 2014 Accepted 20 January 2014
Keywords: Pig Swine Biosecurity Social network analysis Herd health Risk-based disease control
a b s t r a c t A cross-sectional study was conducted between October 2011 and March 2012 in two major pig producing provinces in the Philippines. Four hundred and seventy one pig farms slaughtering finisher pigs at government operated abattoirs participated in this study. The objectives of this study were to group: (a) smallholder (S) and commercial (C) production systems into patterns according to their herd health providers (HHPs), and obtain descriptive information about the grouped S and C production systems; and (b) identify key HHPs within each production system using social network analysis. On-farm veterinarians, private consultants, pharmaceutical company representatives, government veterinarians, livestock and agricultural technicians, and agricultural supply stores were found to be actively interacting with pig farmers. Four clusters were identified based on production system and their choice of HHPs. Differences in management and biosecurity practices were found between S and C clusters. Private HHPs provided a service to larger C and some larger S farms, and have little or no interaction with the other HHPs. Government HHPs provided herd health service mainly to S farms and small C farms. Agricultural supply stores were identified as a dominant solitary HHP and provided herd health services to the majority of farmers. Increased knowledge of the routine management and biosecurity practices of S and C farmers and the key HHPs that are likely to be associated with those practices would be of value as this information could be used to inform a risk-based approach to disease surveillance and control. © 2014 Elsevier B.V. All rights reserved.
1. Introduction The Philippine swine population was estimated at around 1.6 million (M) sows with a total of 12.2 M pigs
∗ Corresponding author. Tel.: +61 467 601055; fax: +61 7 54601992. E-mail addresses: [email protected], ibr [email protected], [email protected] (J.I. Alawneh). 0167-5877/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.prevetmed.2014.01.020
(BAS, 2011). The majority of the pig population is on the islands of Luzon (47%; 5.7 M) and Mindanao (29%; 3.5 M). A key feature of the Philippine pig industry is the smallholder production system (termed smallholder farms). Smallholder farmers (S), as defined by the Philippines Bureau of Agricultural Statistics (BAS) and used for the purpose of this study, were farmers raising one to 41 pigs (young or adult) in their backyards per year (BAS, 2011). Approximately 70% (8.5 M) of the total pig population is found on S
74
J.I. Alawneh et al. / Preventive Veterinary Medicine 114 (2014) 73–87
farms (BAS, 2011), and pigs raised are often used for their own consumption or as financial security (More et al., 1999; Lanada et al., 2005; Lee et al., 2005). Commercial farms (C) as defined by BAS and used for the purpose of this study are farms which satisfy at least one of the following conditions (BAS, 2011): (1) at least 21 head of adult pigs; (2) at least 41 head of grower pigs; or (3) at least 10 head of adults and 22 head of grower pigs. The commercial production system represents 30% of the Philippine pig industry. Smallholder and C farms almost exclusively supply pork to the local fresh meat markets indirectly through livestock traders (Bantugan et al., 1992; More et al., 1999). On a daily basis, livestock traders source their pigs from multiple farms. The daily activity of livestock traders is mainly driven by fresh meat market demands. The structure of the pig industry in the Philippines is likely to increase the frequency of contact (direct and indirect) between herds (Hurnik et al., 1994; Stark, 2000; Ribbens et al., 2009). This increased frequency of contact increases the likelihood and severity of introducing respiratory disease agents irrespective of production type (Gardner et al., 2002; Otake et al., 2010; Lambert et al., 2012b). The consequences of disease introduction were manifested in the porcine reproductive and respiratory syndrome (PRRS) and porcine circovirus type 2 (PCV2) outbreaks in 2008 (Resontoc, 2009). Both S and C farms reported increased mortality in their herds and this resulted in substantial financial and domestic trade implications for the Philippine swine industry (Resontoc, 2009; BAS, 2011). With the silent circulation of infectious agents like PCV2 and PRRS virus, the challenge for regulatory veterinary authorities, farm management and their herd health providers (HHP) is to ensure that sufficient measures (e.g. biosecurity, housing, vaccination, transport, market access) are always in place to reduce the risk of disease introduction into herds (Amass and Clark, 1999; Gardner et al., 2002; Lambert et al., 2012b). There is a paucity of information about farm management, animal husbandry and biosecurity practices of the S and C production systems in the Philippines. Knowledge of ‘weak points’ or areas of ‘vulnerability’ of production systems (‘vulnerability’ can be a subjective assessment of certain management practices such as biosecurity) where infectious pathogens might enter the herd and establish is important because it provides a better focus for disease control and surveillance activities. Filipino pig farmers have access to several HHPs that range in training and experience from agricultural supply store personnel to private swine veterinary consultants. It is plausible to assume that the effectiveness of farm management practices to reduce the risk of disease introduction (or maintenance) in the herd is influenced by farm choice of HHP. Therefore, in a developing country such as the Philippines where financial resources are limited, a better understanding of farm and HHP network dynamics has the potential to form the basis of a risk-based surveillance or disease control programme. Local veterinary authorities could identify ‘vulnerable’ farms and their HHP(s) through routine field investigations. The authorities could then preferentially communicate with HHP(s) of ‘vulnerable’ farms to improve the management practices of the
identified HHPs’ entire client network rather than focusing on individual farms. Social network analysis (SNA; Wasserman and Faust, 1994) offers a means for formalising the process of identifying ‘vulnerable’ farms and their HHPs. Social network analysis provides a network-based approach to reveal the structure of the relationships between farm and HHP (Dube et al., 2009), and to quantify those relationships within the network (Wasserman and Faust, 1994). The use of SNA in animal populations is increasing (Christley and French, 2003; Christley et al., 2005; Martin et al., 2011), and reviews are provided by Newman (2003) and Dube et al. (2009). The objectives of this study were to: (a) group S and C production systems into patterns according to their HHP(s); (b) obtain descriptive information about the grouped S and C production systems in two provinces of high swine density in the Philippines; and (c) identify key HHPs within each production system using SNA. Increased knowledge of the characteristics of S and C farms (e.g. farm type, size, and biosecurity practices) and the key HHPs that are likely to be associated with or influence those characteristics would be of value as this information could be used to inform a risk-based approach to disease surveillance and control. 2. Materials and methods 2.1. Study design and source population A cross-sectional study was conducted between October 2011 and March 2012 in two major pig producing provinces (Bulacan and Pampanga) in Region 3 (Central Luzon). Region 3 was selected because it is one of the most intensive pig rearing regions in the Philippines (BAS, 2011), and has reported major outbreaks and losses from acute respiratory disease in pigs in 2008 PRRS and PCV2 outbreaks (Resontoc, 2009). This study is part of a large abattoir based cross-sectional study that aimed to estimate prevalence and risk factors associated with respiratory disease lesions in finisher pigs at slaughter. Therefore, the target population was pig farms producing finisher pigs in Bulacan and Pampanga provinces. The source population was farms slaughtering finisher pigs on randomly selected calendar dates at government operated abattoirs (n = 29) in Bulacan and Pampanga provinces. Because a complete list comprising source and number of slaughtered finisher pigs was not available prior to slaughter time, systematic random sampling was used at the time of slaughter to select the study population (farms of origin). 2.2. Farm enrolment process Field staff of the Provincial Veterinary Office (PVO) asked meat inspectors at participating slaughterhouses and livestock traders in Bulacan and Pampanga provinces to record the following information at the time that finisher pigs arrived for slaughter: (1) date of slaughter; (2) livestock trader name, address and contact details; (3) livestock transporter name, address and contact details; (4) name and address of the pig farm from which pigs were purchased; (5) date of purchase; and (6) number of pigs
J.I. Alawneh et al. / Preventive Veterinary Medicine 114 (2014) 73–87
purchased. Information provided by the livestock traders to PVO field staff was verified independently by a trained surveyor team (see questionnaire section) and by the named pig farms within two weeks from slaughter date. Farm participation was on a voluntary basis. Because the large study aimed to estimate prevalence and risk factors associated with respiratory disease lesions at slaughter, pig farms were excluded (28%, 184/655; Fig. 1) if there was inaccuracy in the livestock trader’s supplied pig farm details (mainly farm location especially if those farms were located at the borders between Bulacan and Pampanga; 8%, 54/655), pig farms were located in other provinces (4%, 26/655), farms were untraceable (12%, 75/655), or refused to participate (4%, 29/655). Moreover, a farm would also have been excluded (0%, 0/655) if the date of slaughter occurred more than six weeks after the pigs were sold to a livestock trader. A maximum six-week interval was selected based on the fact that respiratory lesions (particularly Mycoplasma hyopneumoniae infection) in pig lungs can resolve eight to 12 weeks after they develop (Kobisch et al., 1993; Fano et al., 2005). 2.3. Farm questionnaire and data management A questionnaire was designed that consisted of three main sections. The first section collected general information about the interviewee (farm owner/manager or person directly working with the pigs on-farm; termed farmer) and farm profiles, such as the farmer demographics and level of education, farm location, type of production system (S or C), farm HHP (semi-open ended question; Table 1), animal husbandry and reproduction management. The second section requested information on the following: whether pigs on the farm come in contact with wild animals or pigs from neighbouring farms, number of workers living on the farm, swill feeding and ‘feedback’ practice (feeding pig manure, viscera or aborted material to pigs to increase herd immunity). Farmers were asked about various farm biosecurity measures: pen hygiene and cleaning frequency, cleaning method and cleaning products used, disposal method for removal of animal waste and faeces, whether traders and their vehicles were allowed to enter the farm and management practices put in place to prevent introduction of disease by vehicles and visitors, ability of farm workers to recognise sick and healthy pigs, animal health monitoring and management, and the interviewees awareness of pig density within a 500 metre radius relative to the farm. The third section asked for information regarding pig management comprising: type of pig housing, source of water supply for the farm and herd size details at the time of the interview, details of pig feed, vaccination routine, and history of any respiratory disease outbreak or outbreak investigation carried out on the farm in the last 12 months. In total, the questionnaire consisted of 71 questions, of which 10 (14.0%) were open, 13 (18.3%) were semi-open and 48 (67.7%) were closed questions with mostly dichotomous answers. Questions (n = 6) with distinct subordinate parts had at least five possible options. The responses were collected through face-to-face interviews conducted by four experienced interviewers
75
(two teams each comprising two interviewers) between October 2011 and March 2012. As there are different dialects in the Philippines, the questionnaire was written in English and translated to the appropriate dialect at the interview. To reduce information bias the questionnaire was pretested on experts in the Philippines pig production systems comprising regional and provincial veterinary officers and animal health advisors. All questions in the questionnaire were clarified with all interviewers before the study start date. The interviewers were instructed to ask questions exactly as stated in the questionnaire and provide only non-directive guidance. To minimise inter-observer variability in conducting the interview, all observers and PVO personnel met after the questionnaire was piloted on the six farms to agree on a common interpretation of the findings. If there was disagreement, the interpretation of the PVO personnel was chosen. To minimise information (misclassification) and selection biases, the interviewees were asked to verify the trader’s identity, date when the pigs were sold and number of pigs sold for slaughter before an interview was conducted (Fig. 1). The validity of the collected questionnaire data was confirmed during follow-up visits to six farms (three in each province) by the first author, the interviewers and Provincial Veterinary Officers personnel. To reduce misclassification bias that could arise from coding errors, the interviewers and the first author checked and corrected impossible coding of categorical variables (n = 80) and unreliable outlier values for continuous variables (n = 3). Throughout the study period questionnaire data were transferred three times a week (Monday, Wednesday, and Friday) to a purpose built relational database (MicrosoftTM Access® 2010, Microsoft Corporation). 2.4. Statistical analysis The unit of analysis was the individual farm. Continuous variables are described using mean, median, minimum, maximum values, 1st and 3rd quartiles, while categorical variables are presented as percentages. Farm profile, farmer demographics, and farm management practice data were described, and the associations with production type were assessed using Fisher’s exact 2 test for categorical variables or Wilcoxon rank sum test to test the equality of the medians for continuous variables. The tests were conducted using R (R Development Core Team, 2012). 2.4.1. Grouping farms based on herd health providers Agglomerative hierarchical cluster analysis (Ward, 1963; Everitt et al., 2001) was used to identify similar subgroups of farms (S and C) ‘homogenously’ clustered based on their HHP(s). Using an iterative procedure, at each stage squared distances between clusters were recomputed by the Lance–Williams dissimilarity update formula according to Ward’s minimum variance method. Ward’s method was selected because clusters are merged based on a minimum increase in total within-cluster variance after merging, therefore resulting in compact and spherical clusters (Ward, 1963; Everitt et al., 2001). To assess the certainty in the clustering, a thousand bootstrap replications were run for each cluster and a probability value
76
J.I. Alawneh et al. / Preventive Veterinary Medicine 114 (2014) 73–87
Fig. 1. Flowchart of farm (smallholder[S] and commercial[C]) enrolment, numbers and reasons for exclusion, number of farms eligible, and farms available for the final analysis through the study period between October 2011 and March 2012in two provinces in the Philippines.
(approximately unbiased (AU) P-value) was generated through multiscale bootstrap resampling. The agglomerative coefficient (AC) was also calculated to measure the clustering structure found by this technique. Both the AU P-value and AC values can range from 0 to 1 with higher values indicative of greater confidence in the clustering structure. The significance of the clusters was set at 0.9 for the AU P-value of the clusters. The AU P-value and AC values were derived using the Pvclust (Suzuki and Shimodaira, 2006) and Cluster (Maechler et al., 2012) packages in R. 2.4.2. Farm and herd health providers networks Five matrices were constructed. In these matrices farmers formed rows (in SNA terminology farmers would be called ‘actors’ and HHPs would be called ‘affiliates’). The interaction between farmers and their HHPs formed the ‘ties’ in the matrices. Ties were considered as undirected and were dichotomised, 1 indicating the presence of the tie and 0 otherwise. Farmers without a HHP (i.e. isolated; n = 3) were identified and removed before further analyses were conducted. Social network metrics, analyses and figures were derived using R, Ucinet 6.421 (Borgatti et al., 2002), and NetDraw 2.123 (Borgatti et al., 2002). A focus of the current study was to explore the network of the farmers and then use that network to identify the main HHP(s) (‘more centrally located’) in the overall network and within each of the identified farmer group network structures. Thus the eigenvector and normalised
eigenvector centrality measures for a 2-mode network were calculated. Eigenvector centrality is a weighted degree measure in which the centrality of a farm/HHP is proportional to the sum of centralities of the HHP/farm it is connected to. The normalised eigenvector centrality is the scaled eigenvector centrality divided by the maximum difference possible in a given network expressed as a percentage (Borgatti et al., 2002). The eigenvector value is a measure of farmer or HHP importance based on their activity, position, and contacts or popularity in the network. Because each farmer could have more than one HHP, it was of interest to determine whether pairs of farmers who engage with one HHP are more likely to also engage with another. To do this, 10 matrices each were constructed for the S and C production systems. Each of these symmetric matrices was specific to one of the 10 HHPs and consisted of rows and columns equal to the number of farmers in the production system. A connection between each farmer pair (of the same production type) was coded as 1. Jaccard’s coefficients were then estimated for HHP matrix pairs using the Quadratic Assignment Procedure (QAP; Hubert and Arabie, 1989) with a permutation-based significance test involving 5000 random permutations. Jaccard’s coefficient, measures the ‘similarity’ of farmer ties (binary) between HHP matrix pairs and is expressed as a proportion where a value of 1 indicates complete similarity and 0 indicates complete dissimilarity. The QAP randomly permutes rows and columns (synchronously) of one matrix
J.I. Alawneh et al. / Preventive Veterinary Medicine 114 (2014) 73–87
77
Table 1 A complete list of herd health providers (n = 10; herd health providers abbreviation used in throughout the manuscript in parentheses) included in the farm questionnaire conducted in two provinces in the Philippines between October 2011 and March 2012. The list was presented to the interviewees (n = 471) as a semi-open ended question with multiple choices allowed. Herd health provider
Definition
Technical consultant contract grower (TCCG)
A veterinarian providing herd health services related to pig health and performance in the growing stage A veterinarian from the government or the private sector contracted to provide herd health service and monitor farm performance at all stages of production cycle A technical-level government employee who implements local and national livestock programmes. These individuals conduct routine rounds in their areas of responsibility to inspect and monitor the health of livestock and extend veterinary services to smallholder farms A veterinarian who resides on the farm as farm manager and provides herd health service at all stages of production cycle An organised group of pig farmers who operate autonomously and privately as a national union or a local cooperative to support member interests in swine production. It has concrete programmes in swine production including credit and marketing assistance for members A government-employed veterinarian who conducts animal disease surveillance (passive surveillance) and investigates reported disease outbreaks in the field Neighbour pig farmer (usually among smallholders), a co-member of the Hog farmer association, or a co-member of the Farmer co-operative An organised group of pig farmers who operate autonomously and privately as a national union or a local cooperative to support member interests in swine production. It has concrete programmes in swine production including credit and marketing assistance for members An individual with technical-level expertise who work for multi-national pharmaceutical companies to promote and sell their products. They provide technical information about their products as well as technical services such as diagnostic services relative to the purchases made by farms with large operations and feed mills A retailer – with or without veterinary or technical qualifications – of agricultural and veterinary supplies such as pharmaceutical drugs, supplements, feed and mixes for livestock who operate in towns and is accessible to farmers. Retailer activities are not monitored nor do they report to any regulatory veterinary authority Other herd health provider not listeda
Private veterinarian consultant (PC) Livestock inspector–agricultural technician (LIAT)
On-farm veterinarian (OFV) Hog farmer association (HFA)
Government veterinarian (GV) Fellow pig farmer (FPF) Farmer co-operative (FC)
Pharmaceutical company representative (PCR)
Agricultural supply store (AS)
Other a
No other herd health providers were listed by farmers.
and recomputes Jaccard’s coefficients in order to compute the proportion of times (per n permutations) that a random measure is larger than or equal to the observed. A low proportion ( 0.90) within each production system (S farm clusters A and B, termed SA and SB respectively; C farm clusters A and B, termed CA
and CB respectively). Descriptive statistics of farm profile, farmer profile, and details of farmers’ choice of HHP(s) according to production type are shown in Table 2. Overall, S and C farmers had different (P < 0.01) combinations of HHP. Smallholder farmers tended to use agriculture supply store (AS; 63%), Livestock inspector–agricultural technician (LIAT; 50%), and government veterinarian (GV; 28%) as their preferred HHP(s), while commercial farms were more reliant on private veterinary services; 61% used private consultants (PC), 51% used on farm veterinarian (OFV) and 48% used pharmaceutical company representatives (PCR). Within S farms, the distribution of farms into the two clusters differed (P < 0.01) between provinces. Compared with SB (n = 275), SA (n = 96) had on average more productive sows (P < 0.01), more pigs on the farm (P < 0.01), were more likely to house pigs in separate pens (P < 0.01) rather than free range (P < 0.01), used a private veterinarian (P < 0.01) as the main HHP, and were less dependent on pig farming as a main source of income (P < 0.01). For C farms, CB (n = 33) had fewer pigs (P < 0.01) and were more likely to use livestock inspectors (P < 0.01), pharmaceutical company representatives (P = 0.02), government veterinarians (P = 0.01), and AS (P < 0.01) as their HHP compared with CA (n = 67). Descriptive statistics of farm management and biosecurity practices are presented in Tables 3 and 4. Compared with S farms, overall, C farms showed higher frequency
78
J.I. Alawneh et al. / Preventive Veterinary Medicine 114 (2014) 73–87
Table 2 Descriptive statistics of farm (n = 471) and farm manager profiles classified by production type (smallholder, n = 371; commercial, n = 100) and herd health provider (smallholder A, SA, n = 96; smallholder B, SB, n = 275; commercial A, CA, n = 67; and commercial B, CB, n = 33) for participating pig farms from two provinces in Philippines between October 2011 and March 2012. Profile
Smallholder farms SA n (%)
Farm profile Province Bulacan Pampanga Number of productive sows Median; Min, Max Mean; 1st quartile, 3rd quartile Total number of animals Median; Min, Max Mean; 1st quartile, 3rd quartile Housing type Separate pens Pen adjacent to house Free range Other Herd health provider None Technical consultant contract grower Private veterinarian (consultant) Livestock inspector–agricultural technician On-farm veterinarian Hog farmer association Government veterinarian Fellow pig farmer Farmer co-operative Pharmaceutical company representative Agricultural supply store Farm manager profile Gender Female Male Education Elementary or cannot read High school College Years of experience Median (Min, Max) Mean (1st quartile, 3rd quartile) Proportion of income derived from raising pigs 75% Unknown
SB n (%)
52 (54) 44 (46)
113 (40) 162 (60)
0; 0, 30 3; 0, 3
0; 0, 18 1; 0, 0
14; 1, 92 22; 4, 11
6; 0, 77 9; 4, 11
61 (64) 32 (33) 1 (1) 2 (2)
120 (44) 87 (32) 61 (22) 7 (3)
2 (2) 4 (4) 72 (75) 40 (42) 4 (4) 1 (1) 1 (1) 1 (1) 1 (1) 6 (6) 14 (15)
1 (0) 3 (1) 0 (0) 145 (53) 4 (2) 0 (0) 103 (38) 61 (22) 3 (1) 8 (3) 220 (80)
28 (29) 68 (71)
84 (31) 191 (69)
11 (12) 45 (47) 40 (42)
25 (9) 154 (56) 96 (35)
7 (1, 39) 10 (5, 12)
7 (1, 42) 9 (5, 9)
39 (41) 19 (20) 8 (8) 2 (2) 28 (29)
79 (29) 95 (35) 2 (1) 1 (0) 98 (36)
Commercial farms a
P
0.03
CA n (%)
S vs. C CB n (%)
24 (36) 43 (64)
18 (55) 15 (45)
223; 0, 2580 395; 121, 425
1; 0, 350 39; 0, 35
1520; 1, 19,630 3102; 629, 3232
22; 0, 6439 481; 6, 382
