"Ghost sows" and the other 8 most frequent errors when processing data (1/2)

Today, virtually all farms use management software for their production data. These softwares are very useful tools, both to facilitate the daily work and to monitor and analyse the farm data. But, to fulfil their role, the type, reliability and processing of the data are critical.

Too often one or more of the following errors are made:

1. Backward data processing. This is one of the most frequent errors and it generates a first problem with reports of job control (worklists, sows records, delayed animal warnings, etc.). This information is valuable if received on time; otherwise it loses much or all of its interest. The situation can be even worse when a problem needs to be assessed and data explaining the situation are not available due to them not having been entered or processed when needed. It is a too frequent and very frustrating situation, both for the farmer and the veterinarian.
 

2. Impossible data processing. The procedure is known as Database Integrity Control. Some software packages allow the entry of improbable or impossible data, probably the most common cause of generation of biased analysis for several variables (farrowing rate, weaned piglets per sow, or per sow per year, non-productive days, etc.).

Obviously, a software that does NOT allow entering this type of information should be chosen. The most common are:

• Gestation > 125 days. Farrowing is virtually impossible after that time, so sows in that situation should be examined and then removed.
• Total born = 0. It is obviously impossible. Often mistaken for born alive = 0, which can sometimes happen.
• Weaning-to-service interval > 65 d. As a general rule, sows are not allowed to have more than 3 oestrus before being serviced. When that happens, the sow is usually a casualty or has already been culled, so it is also a useful filter.
• Age at first service > 400 d. Despite the tendency to service older sows in recent years, 400+ days is not a possible service age, both because of the reasons previously mentioned (the sow is most likely no longer on the farm) and because if the sow is still in the farm, its use as a breeder is not at all recommended.
• Total number of weaned piglets in a sow's productive life > 135. This number is effectively impossible, because in order to achieve it, the sow should have nursed piglets for a minimum of 6 farrowings, and weaned 24 piglets. Such figures usually appear as a result of errors when entering the number of weaned piglets.
• Reason for slaughter ‘anoestrus’ after recording a service. This error rather belongs to a different area, that of the incorrectly-recorded or non-specific death causes, which usually generates as high a degree of confusion that the report is effectively useless. To further complicate matters, even when the causes of death are being properly recorded, each farm uses its own nomenclature (sudden death / heart attack / early death, etc.)

3. Keeping the ‘ghost sows’. This term refers to the sows that are physically removed from the farm (i.e., dead or culled), but not eliminated from the software. From the time of their death, their presence —which is only virtual— continues to build up, thus biasing the calculations based on census or services, farrowings or weanings, depending on the time they are eventually erased from the software. It is strongly recommended to seek and remove them periodically, based mainly on days without activity or days elapsed since a specific event.

Figures 1 and 2 Farm with farrowed or weaned sows for more than 50 days without further events.

4. Farrowing rates of 100% in gilts. In some farms, sows are registered at their first farrowing, fact that would explain the 100% rate. Although the most common situation is to register the sows on their first service, both situations are undesirable as they do not provide any information about the sow's previous period, resulting in a loss of valuable information regarding the acclimatization period, impact of age at first service and even abortions in gilts which, on the other hand, are usually the most affected group for that indicator.