Table 2. Category of use of consumer purchase data (CPD) for food-borne outbreak investigations, 1 January 2006−20 October 2017.

Reference	Purchase data source	Category of use of CPD	Description of use of CPD	Type of outbreak	Type of vehicle
[5] Shah et al., 2009	Loyalty card; 8 cases.	Hypothesis generation and trace-back investigation	CPD used for hypothesis generation/ support of hypothesis and aid in trace back	Dispersed one-province cyclosporiasis outbreak	Organic basil was the most likely vehicle
[6] Gillesberg Lassen et al., 2013	Debit card information, several supermarkets; no. of cases not stated (< 10).	Trace back	Vehicle (frozen berries) found by case–control study; CPD used to identify type and identity of product.	National, later international (4 countries) hepatitis A outbreak	Brand of frozen strawberries sold in (internationally operating) supermarket chain
[7] Collier et al., 2014	Data from membership/loyalty cards from a retailer; no. of cases not stated.	Case finding, trace back, targeted intervention of exposed (information, post-exposure vaccination)	CPD Improved validity of initial hypothesis and targeted post exposure prophylaxis with both hepatitis A virus vaccine and immunoglobulin.	Dispersed national hepatitis A outbreak	Frozen pomegranate arils
[8] Swinkels et al., 2014	Loyalty card purchases in 3-month period; 6 cases.	Trace back	Vehicle identified in part using classical epidemiology, CPD used to locate particular producer and confirm the source. No case–control study done.	Dispersed province-wide hepatitis A outbreak	Frozen berry blend
[9] Hächler et al., 2013	Shopper cards/loyalty cards; 4 cases.	CPD support existing evidence	Supported existing evidence, use delayed by legal clarification. Consent from the patients and the retail company.	Dispersed local listeria outbreak	Cooked ham
[10] Taylor et al., 2012	Loyalty card purchases; 4 cases.	Assists hypothesis generation, trace back	Epidemiological investigation points to vehicle. CPD in subset of cases corroborates and leads to fast trace back.	Dispersed multi-province salmonella outbreak	Ready-to-eat pork product, known as head cheese
[11] Zenner et al., 2014	Till entries and receipts from single restaurant; 41 cases.	Hypothesis generation, analytical study	Helps locate dish on menu in take-away restaurant + makes analytical argument by comparing sale over different time periods.	Point-source (geographical) outbreak associated with single restaurant	Chicken dish, one item of many on a restaurant menu
[12] Bedard et al., 2014	Shopper card purchases, no of cases not stated (< 10)	Hypothesis generation	CPD gives 3 distinct hypotheses, leads to source identification by microbiological testing.	Local county investigation and multi-state cluster	Pine nuts sold in supermarket/stores
[13] Grinnell et al., 2013	Shopper card purchases; 9 cases.	Trace back	Standard epidemiological methods identify vehicle, CPD used to zoom in on producer and exact product.	Dispersed multi-state salmonella outbreak	Industrial chicken products sold in supermarket chain(s)
[14] Routh et al., 2015	Loyalty card purchases; 3 cases.	Trace back	Trace back (helping to identify the vehicle, combined with traditional methods).	Dispersed national salmonella outbreak	Ground turkey
[15] Gieraltowski et al., 2012	Store membership card purchases; 7 cases initially, 19 cases at late stage.	Hypothesis generation (and trace back)	CPD information points to specific hypothesis. Also strongly aids trace back.	Dispersed multi-state salmonella outbreak, 2 serotypes and several vehicles	Salamis made with contaminated black and red pepper (dried spices)
[16] Schneider et al., 2011	Loyalty cards; 11 cases.	Aided trace back	CPD improves validity of questionnaire findings. CPD used to target trace back combined with records of beef processing.	National, multistate salmonella outbreak	Ground beef
[17] Miller et al., 2013	Shopper card purchases; 3 cases.	Trace back	Vehicle suspected by epidemiological methods, small outbreak, evidence in-conclusive. CPD gives GTIN numbers which leads to precise trace back, identifying product.	Dispersed, but small, salmonella outbreak in part of 1 state	Fresh berries, sold in supermarket chain, traced back to specific producer
[18] Müller et al., 2016	Digital receipts from cashier systems from 2 supermarket chains of purchases in 6-week period; 15 cases.	Hypothesis generation	Initial hypothesis-generating interviews are inconclusive, but points to 2 supermarkets. CPD leads to quite specific hypothesis. Followed by traditional case–control study.	Dispersed national salmonella outbreak	Particular type of tomatoes, hidden among all tomatoes in interviews
[19] Ethelberg et al., 2008	Debit cards; digital receipts from several supermarket chains purchases in 6 week period; no. of cases not stated (ca 25).	Hypothesis generation	Many different investigation methods in use. CPD applied on several supermarkets/shops. No common pattern was identified.	Large and prolonged nation-wide salmonella outbreak	Vehicle/source never identified
[20] Bone et al., 2010	Loyalty card purchases three weeks before onset; 9 cases.	Trace back and corroboration of hypothesis	Epidemiological investigation points to vehicle. CPD corroborates (9/9 cases bought product) and points to single brand. Recall without case–control study or microbiological proof.	Dispersed national salmonella outbreak	Dried salami, distributed nation-wide, sold in single supermarket chain
[21] Gossner et al., 2012	Loyalty card purchases; 39 cases.	Trace back and semi-analytical use	Epidemiological investigation points to vehicle. Focused CPD corroborates and points to single brand. Proportions used for likelihood argument. Recall without case–control study or microbiological proof.	Dispersed national salmonella outbreak	Dried salami, distributed nation-wide, sold in supermarket chain
[22] Wilking et al., 2012	Employee cards used for cafeteria sales; 23 cases and 30 controls.	Analytical study	CPD data used for nested case–control study within cohort of company workers	Point-source outbreak, sub-outbreak within large national STEC outbreak	Raw sprouts served as part of lunch meals
[23] Barret et al., 2013	Shopper card purchases; 5 cases (though not clearly stated).	Trace back	Find the exact brand of product after vehicle has been identified using epidemiological methods	Regional (sub-national) STEC outbreak	Fresh ground beef (burgers) sold in supermarket chain.
[24] Ethelberg et al., 2009	Debit cards; digital receipts from purchases from 2 supermarket chains in 6-week period; 7 cases.	Hypothesis generation.	Initial hypothesis-generating interviews are inconclusive, but points to single supermarket. CPD leads to specific hypothesis. Further proof from case–control study and microbiological testing.	Dispersed national STEC outbreak among children	Organic, fermented salami made of beef, distributed nation-wide, sold in single supermarket chain

CPD: consumer purchase data; GTIN: global trade item number; STEC: Shiga toxin-producing Escherichia coli.