“Good” health records: Working toward quality dairy health data

 Resize text         Printer-friendly version of this article Printer-friendly version of this article

Health records should support cow-level health management, herd-level health process management and residue avoidance. The health records on many dairies aren’t “good” and I have the references to prove it! They document a lack of accuracy and consistency needed for efficient summary and analysis. Yet, in the nearly 17 years I have been paying attention, the dairy industry seems to be largely unimpressed and there has been little change. In fact, a survey of Washington and Idaho dairy farmers found that only 17 percent of 160 respondents were dissatisfied with the quality and utility of their computerized health records. However, only 11 percent of 38 Washington and Idaho dairy veterinarians were satisfied with the quality of their client’s health records. Perhaps it has something to do with perspective.

In 1995, Etherington et al. drew attention to the fact that producers generally utilize individual animal data, whereas practitioners and other consultants are more concerned with herd summary information. Since health records are user-defined (i.e., no industry or software standard) and the primary users are dairy personnel, focused on day-today activities involving individual cows, it makes sense they think dairy health records are good even if veterinarians don’t. If we want health data that can be summarized to objectively evaluate the efficacy of health management on client dairies then we need to take a more active role in how health records are kept.

Health records are “good” if they support three critical functions that address the needs of all users:

  1. Cow-level health management
  2. Herd-level health process management
  3. Residue avoidance/regulatory compliance (see sidebar).

In a previous article, I referred to “The 3 Simple Rules of Good Recording” (July 2013) as a guide to implement standard health data-entry protocols on dairies to achieve good health records. This article provides examples of the application of those rules to address health data-management situations that can negatively impact record quality.

Record all disease episodes identified
Regardless of severity, duration or treatment given, all identified disease episodes should be recorded. For the most accurate estimate of disease incidence (prevention efficacy), standard disease case definitions need to be adopted. In 1998, Kelton et al. proposed recommended standards for recording and calculating common clinical diseases of dairy cattle, yet little progress seems to have been made. Recording of metritis and mastitis is particularly challenging, primarily because case definition often includes treatment-decision considerations.

It is relatively straightforward to define metritis as an abnormal uterine discharge (i.e., foul smelling and watery) in a post-partum cow. Whether or not she has systemic signs such as pyrexia is a qualifier that may further define the severity of the clinical episode and often determines if intervention (e.g., antibiotic treatment) is warranted. Similarly, mastitis can be simply defined as abnormal milk (i.e., flakes, clots or watery) likely associated with infection. The degree of milk abnormality or systemic involvement, again, defines severity and commonly serves as criteria for treatment decision making.

Censoring cases that don’t warrant therapy presumes we know they don’t have an impact on health and productivity worthy of attention. For example, maybe we assume that a cow with a flake or two in the milk is likely going to be no growth on milk culture and doesn’t need treatment. However, by recording those episodes maybe we find 40 percent have a more severe clinical episode within two weeks that is treated. Does that mean they should all be treated on that first flake? I don’t know, but if recorded, we, as the health management professionals on the dairy team, can evaluate the epidemiology and economics of such cases and provide more evidence-based prevention and treatment protocol recommendations.

I am optimistic that industry-wide agreement on the minimum clinical signs that define a disease can be achieved quickly through a cooperative, concerted effort. In the meantime, lack of standard disease definitions should not preclude accurate and consistent recording of disease, whatever the definition. Most clients want to know how they compare with their peers. Capitalize on the desire to benchmark to achieve greater consistency in disease definition and treatment among your clients (see “Recommended Disease Definitions” at goodhealthrecords.com).

Treatment decision making has been the greater challenge thwarting adoption of standard disease definitions. From the veterinarian to the calf doctor, each has pride and strong conviction that his/her treatment protocols are “the best.” Perhaps this is why in a survey of over 200 dairies we found 11 different drugs besides intramammary antibiotics being used to treat mastitis. Sixty percent of uses were extra-label; 7 percent were prohibited. Metritis was treated with 10 different antibiotics; 85 percent of uses were extra-label and 7.6 percent were prohibited. Is so much variety truly medically necessary? I believe that this “creative” drug use is fostered by misconceptions of the uninformed. What is the quality of evidence you have for your treatment protocols? Accurate and consistent health records and the ability to efficiently evaluate them would allow informed decision making based on evidence rather than perception.

Accurately record “no treatment”
Too often if a cow does not receive a treatment that has a withdrawal, the fact that they had clinical disease isn’t recorded. Or, if the disease is recorded, the fact that the case was not treated is not noted. Typically “no treatment” refers to a decision not to treat a clinical disease episode with antibiotics.

This is most commonly encountered in herds that make clinical mastitis treatment decisions based on milk culture results. Those clinical episodes that come back as no growth or Gram negative (coliform) are not treated with an intramammary antibiotic. The first step toward accurate and consistent health-data recording is to record the fact that no treatment was chosen. If treatment is recorded as a two-character remark, such as SP — spectramast or TD — today, then NT — no treatment should be as well. This practice also keeps disease remark entries consistent (the third simple rule of good recording).

“No treatment” may also be elected when it is decided to remove a cow from the herd as a result of a clinical disease episode. Two common data-recording problems are encountered in this situation. The first is when the clinical episode that resulted in removal is never recorded. Obviously, this results in censoring of clinical cases with arguably the worse outcome. The second involves interpreting the outcomes of no treatment. We are commonly asked to compare the outcomes between cases that receive no treatment versus antibiotic treatment. To do this accurately we need to know whether a no treatment decision was with the intent of keeping the cow or removing the cow. The intent can be communicated effectively using two different remark abbreviations (e.g., NT — no treat, intend to keep, BF — no treat, intend to remove). This allows exclusion of those cows that were not treated because they were to be removed from the analysis. Of course, such a comparison is predicated on the assumption that treatment and no treatment were applied without case-selection bias, which is seldom true. Nonetheless, differentiating the intent of no treatment is important for evaluating adherence to established treatment protocols. Accurate and consistent health-data recording fosters maintenance of a valid veterinary-client-patient relationship (VCPR) when you routinely evaluate the distribution of treatments.

We often find that the “lesion location” (foot for lameness, quarter for mastitis) is not recorded when “no treatment” is selected for a clinical case. This is problematic if you want to track clinical episode outcomes at the quarter or foot rather than the cow-level. For example, a cow has a clinical mastitis episode in the right rear quarter treated with intramammary antibiotics recorded. A month later the same cow has a clinical mastitis episode that received no treatment and the quarter is not recorded.

Was the second episode also in the right rear (suggesting possible treatment failure the first time or a damaged teat end) or a different quarter? Lesion location is not recorded if treatment isn’t given when our mindset is solely on tracking drug use.

Use a single, specific term to record clinical disease
Health data-entry protocols should be established to differentiate clinical disease episodes from subclinical disease, identified by screening tests and prophylactic treatment of at-risk cows. We often find herds performing fresh-cow milk cultures will record treated culture-positive cows using the same term that is used to denote clinical mastitis (e.g., MAST). Fresh culture-positive cows that are treated should be recorded with a different term (e.g., FMAST).

Post-partum uterine health-recording practices are commonly focused on recording that a treatment was given rather than that disease occurred. Thus, when summarized, they often do not accurately represent postpartum uterine health. We see three common problems highlighted by these examples:

1) Cows that have a retained placenta recorded at three or more days in milk with a remark indicating antibiotic treatment. These cows usually have clinical metritis (abnormal uterine discharge at a minimum) and have retained fetal membranes. If retained placenta was not previously identified and recorded, a more accurate representation of this case would be to record both retained placenta and metritis. In this case, the antibiotic treatment should be associated with the metritis.

2) Retained-placenta cases at one to two days in milk are recorded as metritis so an antibiotic treatment can be given. This is most commonly encountered in herds that utilize standard treatment protocol functions (e.g., “Protocols” in Dairy COMP 305 and Rx-Plus module for DHI-Plus). In such cases, the defined retained-placenta protocols are not associated with antibiotic treatment. Antibiotic treatment and proper withdrawal times are desired so a metritis that is associated with an antibiotic protocol is recorded. If you approve antibiotic treatment of cows diagnosed with retained placenta, define a treatment protocol for it in the computer so diseases and their management can be recorded accurately.

3) Cows considered to be at-risk for metritis (i.e., those with a dystocia, twins or stillborn calf) are recorded as metritis at zero to two days in milk so an associated antibiotic treatment protocol can be implemented. Instead, treatment of at-risk cows should be recorded using a different term (e.g., RMETR) than that used to denote a cow with clinical metritis (e.g., METR).

The blue graph shows the typical pattern seen when true metritis episodes are recorded as “metritis.” The graph in red shows the pattern observed when at-risk or retained-placenta cows are recorded using the same term used to denote clinical metritis.

Evaluation of the DIM distribution of the term used to record metritis is an easy way to better understand postpartum treatments on the dairy. In the accompanying figure, the typical pattern seen when true metritis episodes are recorded as “metritis” is shown in blue. The graph in red shows the pattern observed when atrisk or retained-placenta cows are recorded using the same term used to denote clinical metritis.

Daily cow-level activities on the dairy generally do not necessitate such detailed recording. When veterinarians are more engaged in routine, herd-level health process-management activities they become critical. And health records that adequately address these needs usually meet requirements for residue avoidance/regulatory compliance. Work with your clients to establish standard disease definitions that will allow comparison among herds. Establish standard health data-management protocols to ensure accurate, consistent recording of those diseases. Finally, use those quality data to generate productive knowledge based on evidence to prove your value as the health management professional on the team and keep your VCPR valid. If you want help, visit goodhealthrecords.com or email me at jrwenz@vetmed.wsu.edu.

Good records allow more evidence-based prevention and treatment protocol recommendations.


Prev 1 2 3 4 5 Next All



Comments (0) Leave a comment 

Name
e-Mail (required)
Location

Comment:

characters left


Sil-All®

With feed values near record levels and potential shortages of top-quality feeds, Sil-All® provides a cost-effective option to help control ... Read More

View all Products in this segment

View All Buyers Guides

Feedback Form
Leads to Insight