Cumulative pregnancy by DIM

Definition

Cumulative pregnancy by DIM is the share of cows that have become pregnant by each point in lactation, plotted as a curve against days in milk. It starts at 0% at the VWP and rises with each 21-day cycle. It is the “time to pregnancy” view of reproduction: the shape of the curve shows how fast cows get pregnant, and where it flattens shows how many never do.

The headline point is % pregnant by 150 DIM: of the cows that calved, how many conceived by 150 days in milk. It folds heat detection, conception and pregnancy loss into one number that is easy to compare between herds and over time.

Done properly it is a survival (time-to-event) curve, built from a cohort of calvings:

Each cow enters at calving.
She leaves the at-risk group when she conceives (an event) or when she is culled, dies, or is marked do-not-breed (censored, not counted as a failure or a success).
At each DIM, cumulative % pregnant = 1 − the share still not pregnant among cows still at risk.

This avoids the survivorship bias of averaging days open over pregnant cows only.

% open past 150, 200 and 270 DIM

The mirror image is a snapshot of today’s herd: how many cows past a DIM cutoff are still not pregnant.

% open > 150 DIM: cows over 150 DIM that are not pregnant ÷ all cows over 150 DIM. The early warning: these cows already have a long days open.
% open > 200 DIM: cows that are likely to need a decision about whether to keep breeding them.
% open > 270 DIM (or the herd’s DNB cutoff): cows that in most herds should already be marked do-not-breed or culled. A high number here usually means DNB is not being recorded.

Count open, bred-not-yet-checked and fresh-not-yet-bred cows as “not pregnant”, but show them separately: a bred cow at 160 DIM may turn out pregnant. Leave DNB cows out of both numerator and denominator, and report how many there are.

Formula

Plain text (life-table, by 21-day DIM bucket after the VWP):

Pregnancies in bucket k = cows that conceived with DIM at conception in bucket k
At risk in bucket k     = cows not yet pregnant and still in the herd and breedable at the start of bucket k
                          (count cows censored partway through as half, by convention)

Bucket PR_k = pregnancies in bucket k / at risk in bucket k

Cumulative % pregnant by end of bucket K = (1 − Π_{k ≤ K} (1 − PR_k)) × 100

% open > X DIM (snapshot) = cows with DIM > X not pregnant / all cows with DIM > X × 100   (DNB excluded)

LaTeX:

S(K) = \prod_{k \le K} \left(1 - \frac{P_k}{R_k}\right)
\qquad
\mathrm{CumPreg}(K) = \big(1 - S(K)\big) \times 100
\qquad
\%\mathrm{Open}_{>X} = \frac{N_{\mathrm{DIM}>X,\ \text{not preg}}}{N_{\mathrm{DIM}>X}} \times 100

Where P_k is pregnancies and R_k cows at risk in DIM bucket k. With 21-day buckets starting at the VWP, the bucket PRs are the herd’s pregnancy rate by stage of lactation. As a rough check: at a constant 25% PR with a 60-day VWP, 150 DIM is about 4.3 cycles past the VWP, so about 1 − 0.75^4.3 ≈ 71% of cows would be pregnant by 150 DIM before pregnancy loss and culling.

How to pull it in DC305

DC305 has no single standard command that draws the full survival curve for every version. Snapshot percentages come straight from PCT:

PCT DOPN<150 DCC>0 FOR LACT>0                % of cows that conceived before 150 DIM
PCT DCC=0 DIM>300 FOR LACT>0                 % of cows open past 300 DIM
COUNT ID FOR LACT>0 DIM>150                  all cows past 150 DIM
COUNT ID FOR LACT>0 DIM>150 RC<5             of those, not pregnant (DNB, fresh, open, bred)
COUNT ID FOR LACT>0 DIM>150 RC=1             of those, do-not-breed

RC codes are 1 DNB, 2 fresh, 3 open, 4 bred, 5 pregnant, 6 dry (see RC codes). DNB is RC 1, so RC<5 includes DNB cows; subtract them. Repeat with 200 and 270 for the other cutoffs. PCT, COUNT and LIST cover live animals only by default, so these snapshots miss cows that were culled open; \B adds animals that have left the herd (sold or died) and \D gives those alone.

For the cohort curve:

Export cows that calved in a window (for example 12–18 months ago, so that most have reached 150+ DIM) with fresh date, conception date and leave date, including cows that have since left the herd (\B). Then build the life table outside DC305.
Or use a graph: GRAPH has a survival-curve switch (\S).

Typical range

Herd context: US Holsteins, lactating cows, freestall or dry-lot, VWP 50–60 days, mostly timed AI with some detected heats.

Measure	Typical	Good
% pregnant by 150 DIM (cohort)	55–65%	> 70%
% open > 150 DIM (snapshot, DNB excluded)	20–30%	< 15–20%
% open > 200 DIM	10–15%	< 10%
% open > 270 DIM (not DNB)	3–5%	near 0%

The curve should rise steeply in the first 2–3 cycles after the VWP. A curve that is flat for the first 30 days past the VWP points to slow first service (see days to first service). A curve that rises early and then flattens below 80% points to poor resynchronization or re-breeding of open cows.

Seasonal calving cohorts differ: cows calving just before summer heat stress usually have a lower % pregnant by 150 DIM in hot regions.

Common mistakes

Using a snapshot as if it were the cohort number. ”% of cows over 150 DIM that are pregnant” today leaves out cows culled open, so it reads high.
Counting culled or DNB cows as open (or as pregnant). They should be censored in the cohort curve and left out of the snapshot.
Using too young a cohort. Cows that haven’t reached 150 DIM can’t contribute to ”% pregnant by 150 DIM”. Pick calving dates far enough back, or use the survival method so they are censored.
Counting bred-not-checked cows as open. Some of them are pregnant. Show them as their own group.
Ignoring pregnancy loss. A cow confirmed pregnant at 120 DIM who later aborts is not pregnant at 150 DIM on a later run. State whether the measure uses first confirmation or current status.
Comparing herds with different VWPs or check timing without saying so.