A total of 95.7% (n = 242) of all study farms in N, 98.8% (n = 249) in E, and 89.2% (n = 232) in S participated in DHI testing. Most of the participating farms always reviewed the monthly test reports to assess their udder health and milk yield data and to be able to react to problems if necessary (N: 90.5%; E: 87.2%; S: 97.4%) (Table 4).

A general use of a VHHC took place in the regions N, E and S at 54.2% (n = 137), 59.9% (n = 151), and 18.1% (n = 47) of the farms, respectively. VHHC specifically for udder health was used by 27.7% (n = 70) of the farms in N, 50.0% (n = 126) of the farms in E, and 7.7% of the farms in S (Table 4).

In case of clinical mastitis, 80. 2% (n = 203), 81. 8% (n = 206), and 73. 1% (n = 190) of all farmers in N, E, and S took milk samples at least occasionally. The farmers declared that milk samples for microbiological analysis were never collected from cases with an increased SCC in 44.7% (n = 113), 40.5% (n = 102), and 51.5% (n = 134) of the farms in N, E, and S, respectively. In all three regions, milk samples were rarely routinely tested before cows had been dried off and about 70% of the farms never took a milk sample for microbiological examination before drying off (Table 5).

Farms using conventional milking equipment carried out pre-milking in over 90% of all cases in all regions. In N (77.3%) and S (58.7%), most of these farms practiced pre-milking onto the floor, and in E, pre-milking onto the floor (47.3%) and into a pre-milking cup (45.9%) were practiced almost to the same extend (Table 5).

The participating farms used different data sources to report the clinical mastitis (CM) incidence (Table 6). In total, 65.2% (n = 165) of farms in Region N, 49.2% (n = 124) of farms in E, and 59.6% (n = 155) of farms in S solely estimated case numbers of mild clinical mastitis cases. Document-based estimation was the source of reported mild CM cases on 44 (17.4%), 36 (14.3%), and 34 (13.1%) of farms in N, E, and S. A herd management program or other own documentation was used on 20 farms in N (7.9%), 61 farms in E (24. 2%), and on 31 farms in S (11. 9%). The farmers documented the severe mastitis cases in a similar manner, even though a fairly large proportion of them stated that they only estimated the number of cases [N: 65. 2% (n = 165); E: 53. 4% (n = 136); S: 61. 2% (n = 159)].

The reported annual median farm incidence of mild clinical mastitis was 14.8% with IQR (interquantile range; Q0.1–Q0.9) 3.5–30.8% in N, 16.2% (IQR 1.9–50.4%) in E, and 11.8% (IQR 0.0–30.7%) in S (Table 7).

The reported annual farm incidence of severe clinical mastitis (cases with disturbance of general condition) had a median of 4.0% (IQR 0.0–12.2%) in N, 2.0% (IQR 0.0–10.8%) in E, and 2.6% (IQR 0.0–11.0%) in S (Table 7).

Most study farms participated in DHI testing. However, in Bavaria, the study region with the smallest farms, the participation frequency in DHI testing was lower than in the two other regions.

DHI testing is an important tool to evaluate udder health (16, 21, 46, 47). Especially SCC data are well suited to evaluate udder health problems on farms professionally (48). It is important to collect as many and as accurate data as possible from the dairy herds (46, 49, 50). According to Schmidt and Smith (51), farms participating in the DHI showed significantly better udder health and higher milk yield than farms which did not take part in DHI testing.

In our study, the proportion of farms participating in DHI testing was noticeably higher (94.5%) than for Germany as a whole (67.5%) (30). It can be assumed that farms participating in DHI testing are also generally more willing to receive external advice. As the percentage of farms participating in DHI testing in our study was higher than the German average, our study population probably represents those herds more willing to receive external consulting. Compared with other countries, it is striking that there was a high degree of willingness to participate, even though regional variations existed, with only 52.2% of herds participating in DHI testing in Thuringia as opposed to 66.6% in Bavaria and 86.6% in Lower Saxony (30). Internationally, highest values were found in Norway (98%) (52). In Canada, 70% of dairy farms participated (52) and in the USA, the participation rate in a DHI program was under 50%, depending on the region even lower, and farms often participate irregularly (47). It would be desirable that a DHI participation rate like the one found in our study would become the German and international standard.

From clinical experience, it was suspected that documentation of clinical mastitis cases on German dairy farms is often incomplete, and this was confirmed. A total of 58.0% of all study farms simply estimated the case numbers during the interview. Electronic herd health and production management programs were rarely used to document and evaluate udder health data (14.6%), and striking regional differences, presumably due to the farm structure and size, could be found (N: 7.9%, E: 24.2%, S: 11.9%). Systematic, detailed and chronologically traceable documentation is the starting point for optimal management and continued monitoring of animal health (53), especially udder health. Furthermore, it also enables data-based cooperation between farmers, veterinarians and consultants. Electronic herd health and production management programs are an important step to optimize dairy farm performance (54).

In summary, other studies confirm that mastitis control measures and good documentation of mastitis are the basis for effective improvements in all mastitis surveillance measures and can improve udder health (55) and economic efficiency in the long term (43).

In our study, VHHC was used on many farms in the North (54.2%) and East (59.9%), while only 18.1% of the farms in the South used VHHC. Regarding udder health, even lower participation rates could be documented in our study (N: 27.7%; E: 50.0%; S: 7.7%). However, it seems that larger farms are more likely to use this. A constant evaluation and use of external consultants can verifiably contribute to an improvement in animal health in dairy herds. Farms that used a VHHC in Switzerland did not have higher veterinary costs per cow, but better herd health for the mainly investigated fertility parameters (56). This was true regardless of herd size, although larger farms often showed a greater self-motivation to participate. In particular, as part of intense VHHC, more comprehensive udder health programs and surveillance systems need to be implemented (57). A joint analysis of the DHI data between the farmers and the veterinarians/advisors can lead to an objective on-farm assessment of the individual udder health situation and identify aspects for optimization (32). VHHC should therefore be used even more frequently in the future as it will become more important for modern dairy farmers to work professionally and effectively in order to compete economically, and it also contributes to continuous development (54). Our findings are consistent with those of Nielsen and Emanuelson (58) who found that loose housing farms and larger farms are more likely to ask for professional advice. There are only a few studies documenting the use of VHHC, but in some countries, such as Denmark, VHHC is widely used, also because it is mandatory for farms with more than 100 cows (59).

In the study population, almost half of the farms (45.6%) never took milk samples for microbiological analysis in the case of elevated cell counts. Even in the case of clinical mastitis, up to 21.3% of the farmers did not take milk samples at all. Before drying-off, the proportion of those who never tested a milk sample was also very high, 73.1% in the N, 69.4% in the E and 71.9% in the S. However, evidence-based treatment of udder infections should be targeted against the causative pathogens, and microbiological investigations of milk samples contribute to a prudent use of antimicrobials in dairy herds (33). In 30% of clinical mastitis cases, no pathogen was detected and quarters with culture negative samples should not be treated with antimicrobials (60). In order to establish an effective standard protocol for the treatment of mastitis, it is first necessary to develop a system for the routine identification and analysis of clinical mastitis cases; this includes the microbiological examination of milk samples (61). Therefore, regular collection of quarter milk samples is highly desirable (62).

The European Medicines Agency (EMA) categorizes antibiotic substances for human medicine into groups of increasing importance. As result there are restrictions in the use of antibiotics in mastitis treatment (63). In Germany, the Regulation on Veterinary Home Pharmacies (TÄHAV) stipulates situations when antibiograms are mandatory (64). Therefore, there are also various legally binding cases in which it is important to analyze milk samples regularly and comprehensively. The prudent use of antibiotics and the reduction in antibiotic use in dairy cows as a whole is therefore an important goal, and in adult dairy cows, the antibiotic doses used relating to udder health management have the highest reduction potential (34). In other studies, it has been proven that screening milk samples, for example with on-farm tests, can significantly contribute to a reduction in the use of antibiotic doses (65, 66). Overall, the potential has not yet been sufficiently used on the study farms.

More than 90% of the farms with a conventional milking system had adapted pre-milking as part of their milking routine. It is optimal to use a pre-milking cup for examining of the secretion, which was only rarely the case on the farms in region N and S. In contrast, usage of a pre-milking cup in region E was practiced to the same extent as pre-milking on the floor. As Vieira et al. (67) and Huijps et al. (68) already stated in their studies, pre-milking is an important visual control point of the milking process for detecting CM. Furthermore, pre-milking does not only activate the neuro-hormonal reflex chain of the cow that is important for milk ejection, but also enables fast and complete milking (69, 70). Pre-milking and using a pre-milking-cup can reduce the rate of new infections (21) and lead to a reduction in SCC (71). In the present study, a higher proportion of farms practiced pre-milking than in the investigations of Dufour et al. (72), where 53% of the farms adapted pre-milking to their milking-routine. It is recommended to pay more attention to this important control point of good milking practice on all farms and during every milking process (68).

Clinical mastitis (CM) in our study was divided into mild and severe cases. Often only one value for clinical mastitis is found in other studies, so the figures are not directly comparable with the present study. The reported annual median incidence of mild clinical mastitis on the farms was 14.8% (N), 16.2% (E), and 11.8% (S). This is significantly lower than, for example the results of Santman-Berends et al. (73) for the Netherlands. However, the datasets in our study are only based on the farmers’ declarations. The incidence of severe clinical mastitis detected in our study (N: 4.0%, E: 2.0%, S: 2.6%) is comparable to the results detected by Verbeke et al. (74), whereas quarter cases per 10,000 cow days were considered in their study.

In our results, it is noticeable that there are large differences between the farms with the lowest and those with the highest incidence of severe clinical mastitis cases as well as mild clinical mastitis cases (Table 7). A wide range between farms like in this study was also found by Olde Riekerink et al. (75) on Canadian farms. The udder health situation in a herd depends on many factors such as hygiene (13, 76), herd SCC (77, 78), parity (76), and milk yield (77). Since we visited over 750 farms with a large variety of production conditions, the observed scatter is not surprising. Overall, incidence of mild mastitis cases can be compared with other results, which range from an average of 14.4% in Austria (76) to 19% in Finland (79), 23 and 23.7% in Canada (75, 80) to 25.7% in another German study (81) and 27.1% in US indoor holdings (82). However, differences in sampling of data or calculations were present between these studies.

We decided to ask farmers to report the cow level incidence because we expected the definition of a new case within the same lactation to differ between farmers. Due to the anticipated low documentation level, we additionally did not expect farmers to be able to distinguish, retrospectively over 1 year, between a second case in the same quarter and a new case affecting a new quarter. Still, in some instances, farmers might have counted more than one case of clinical mastitis per cow and lactation. Due to the large percentage of farmers estimating the number of new clinical mastitis cases, over- or underreporting cannot be ruled out. As it turned out in the research conducted by Grieger et al. (83), clinical mastitis cases often occur as a recurrence; 6–44% of cases are relapses. This fact could have an impact on the data farmers reported to us.

Internationally, a threshold of 200,000 cells per milliliter milk is frequently used [i.e., (57, 84, 85)]. There are publications, which propose that the 100,000 cells per milliliter milk threshold is more sensitive, but not as specific as the threshold of 200,000 cells (86). A threshold value of 200,000 cells per milliliter better reflects the cell count of an infected quarter, but healthy udder quarters usually have a cell count of 70,000 cells per milliliter or less (85, 87–89). Other publications like Adkins and Middleton (89), Krömker et al. (32, 36) and Middleton et al. (90) used the 100,000 cell per milliliter threshold, too. We chose a threshold of 100,000 cells per milliliter milk for our investigations, since this is a frequently used and recognized limit in German-speaking countries and used in the DLQ guidelines. A high percentage of cows with healthy udders guarantees higher milk quality (91), reduces the infection risk, and is the proven threshold which the participating farmers must and can work with. In Supplementary Table 3, all indicators are also reported with a 200,000 cells per milliliter milk cut-off.

The percentage of aWIM found in our study was approximately 60% in all regions. When comparing our results with other publications (32, 36, 57, 84, 85, 89, 90), it is noticeable that the proportion of animals considered to be in good udder health is at a comparable level, despite our stricter SCC limit. With a threshold of 100,000 cells per milliliter, Klocke et al. (35) found a slightly lower average of 54.8% on 48 of the largest dairy farms in Lower Saxony, though the range was comparable between both studies.

The aNIR during lactation on the farms evaluated in this study was 17.1–19.9%, similar to the 19% reported by Krömker and Volling for Lower Saxony (32). Klocke et al. (35) documented an NIR of 22.4% in a study population of a total of 5% of the dairy farms in Lower Saxony. Fauteux et al. (92) documented a new infection rate of 11%, which is almost identical to our results with a threshold of 200,000 cells per mL. However, even within the farms studied, our results showed a large difference between Q0.1 and Q0.9 of nearly 20%. Other publications document similarly high ranges (32, 35).

In the present study, the percentage of aLCC (N: 0.9%, E: 1.1%, S: 0.4%) was similar or even slightly lower than in other studies with 1.8 and 1.7% (32, 35). A single cow with a highly elevated cell count can have a large effect on the cell count of the collected bulk tank SCC, especially in smaller herds. It is readily understandable that the value was somewhat higher in E with significantly larger herds on average, as farmers might be less concerned by individual animals influencing the bulk tank SCC. In contrast, on very small farms, as they were most often found in the region S, the individual animals quickly had a serious effect on the quality of the milk. Thus, such animals are presumably culled more quickly when farmers see only a LCC. Our results are almost identical with the long-term results of the German Dairy Herd testing (46). According to Degen et al. (37), animals with a chronically high SCC are a risk to the udder health of the herd and a high cost factor. Therefore, their proportion must be kept as low as possible.

The HMR showed very large differences in the median between the regions. While in S, 23.5% of cows in first lactation suffered from mastitis, in N, this proportion was 28.4% and in E 35.7%. Our study in S revealed comparable values to those of Bludau et al. (93), who documented 18–27.5% mastitis cases in heifers in Switzerland. Krömker and Volling (32) found a rate of 39% in 84 randomly selected dairy herds in Lower Saxony, Germany, which is comparable to that of our farms in region E. Overall, a wide spread of values and thus a clear potential for improvement can be observed. Especially first lactating cows should have healthy udders at the beginning of lactation in order to achieve a high lifetime performance and reach a higher age (94). However, a high SCC in the herd seems to be one of the most important factors, which can lead to an increase in heifer mastitis (95, 96).

In order for farmers to better assess and interpret their own udder health data, they need to know which factors influence it. For example, the season clearly affects the SCC (92, 97–99). According to Fauteux et al. (92), the season and the associated climatic conditions influence the NIR, this increasing significantly in summer, while herd size and average milk yield have no influence thereupon. The seasonal effects found in this study (Figures 2–4) are consistent with the findings of various other studies; there is usually a peak in the SCC in the warm summer months (92, 97, 99). In our study, the proportion of cows with a cell count of less than 100,000 dropped drastically in the months of July to September. The risk of new infection increases due to a much higher bacterial infection pressure in summer (98). It has been shown that heat stress endangers animal health and this is reflected, for example, in an increase in the individual SCC (100). Also the housing conditions can interact with the effect of season on the SCC in dairy cows. However, it was not the aim of this study to explore this relationship further.