(Received 17 September 2004; accepted 1 July 2005) dairy cow / mastitis / risk / dry-cow treatment / meta-analysis

Transcription

1 Vet. Res. 37 (2006) INRA, EDP Scences, 2005 DOI: /vetres: Orgnal artcle Incdence of ntramammary nfectons durng the dry perod wthout or wth antbotc treatment n dary cows a quanttatve analyss of publshed data Auréle ROBERT a,b *, Henr SEEGERS a, Nathale BAREILLE a a Unt of Anmal Health Management, ENVN/INRA, BP 40706, Nantes Cedex 03, France b ARILAIT Recherches, 42 rue de Châteaudun, Pars Cedex 09, France (Receved 17 September 2004; accepted 1 July 2005) Abstract The study was amed at summarsng the lterature that compares the ncdence levels of spontaneously occurrng ntramammary nfectons (IMI) durng the dry perod, wthout versus wth antbotc dry cow treatment (DCT). A meta-analytc relatve rsk (RR) calculaton was mplemented when a pror relevant. Two man categores of comparson were used n the 36 selected papers. In the frst category, the udder quarters were randomly allocated (at quarter or cow level) to an untreated or a treated group. Quarter ncdence averaged 12.8% (weghted mean) n untreated quarters, and dependng on the DCT used, from 6.6 to 8.0% n treated quarters. The meta-analytc RR of new IMI for untreated versus treated quarters vared from 1.54 to 1.94, dependng on the DCT used. DCT was manly found effectve aganst IMI due to streptococc and coagulase-postve staphylococc. Based on only a few papers, the applcaton of an nternal teat sealer was assocated to a qute smlar (or possbly better) protecton aganst IMI than DCT, but only n a subpopulaton of partcular (selected) cows. In the second category of studes, a selectve dry cow or quarter antbotc-treatment (selectve DCT or DQT), accordng to cow or quarter selecton crtera, was compared to blanket DCT. The meta-analytc RR of new IMI was 1.71 for selectve DCT versus blanket DCT. Selectve DQT seemed to be more at rsk than selectve DCT, but conssted of treatng a much lower proporton of quarters. The summary-results provded by our meta-analyss should only be used wth cauton, due to possbly low external valdty. More research seems to be relevant on the rsk factors of new IMI durng the dry perod to make the outcomes of omsson of DCT n selected cows more predctable under feld condtons. dary cow / mastts / rsk / dry-cow treatment / meta-analyss 1. INTRODUCTION The dry perod (cessaton of regular mlkng) n the dary cow s a senstve perod for the occurrence of new ntramammary nfectons (IMI), especally shortly after dryngoff and before calvng [50]. Snce the 1960s, n order to control the rsk of new IMI after dryng-off, and also to try to elmnate exstng IMI, a blanket dry cow antbotc treatment (blanket DCT) has been recommended at dryng-off for all cows, as a component of the 5 Ponts Mastts Control Plan [19, 36, 49, 51]. Blanket DCT s today wdely mplemented, wth an adopton rate rangng from 75 to 99% dependng * Correspondng author: arobert@vet-nantes.fr Artcle publshed by EDP Scences and avalable at or

2 26 A. Robert et al. on the country [17], except n the Nordc European countres 1. Blanket DCT has been challenged for a long tme by some researchers, holdng that only nfected cows must be treated [17, 21, 25, 38, 72] or due to ts hgh cost [87]. More recently, blanket DCT has agan become more challenged n several countres for multple reasons. The frst one s the consumers and socety s expectatons regardng antbotc resdues n mlk, or antbotc resstance [3, 4, 8, 10]. The DCT s, reportedly, a possble vector of pathogens nducng a new IMI, under certan crcumstances [23, 81]. Moreover, the systematc use of DCT at dryng-off possbly seems no longer necessary, n vew of the reducton of the prevalence of udder-born contagous pathogens, as shown by the mprovement n SCC levels n almost all countres durng the last 20 years [33, 41, 64]. An alternatve, to the strctly preventve use of DCT, s to apply an nternal, or external teat sealer 2 [3, 4, 27, 35, 42, 44, 45, 46, 73, 88]. These sealers can be used n combnaton wth DCT to obtan a preventve and a curatve effect [26, 45, 46, 47, 68, 84]. Another alternatve to blanket DCT s selectve DCT. In ths strategy, only cows or quarters assumed nfected are treated. Selecton of quarters treated can be done at the cow level or at the quarter level. In most countres, these alternatve strateges are not yet wdespread. One explanaton could be the fact that ther economc outcomes for dary farmers are not wellknown or predctable. To allow economc calculatons to compare strateges for DCT (e.g. blanket DCT versus complete omsson or selectve DCT, or use of a teat sealer alone or n combnaton), nformaton s 1 Ekman T., Osteras O., Mastts control and dry cow therapy n the Nordc countres, Proc. Annual Meetng Natonal Mastts Councl, Lansng, Mchgan, USA, 2003, pp Tmms L.L., Effcacy of barrer teat dps n preventng dry perod mastts, n: Smth K.L. (Ed.), NMC, Proc. Natonal Mastts Councl Regonal Meetng, Syracuse, NY, USA, 1997, pp needed about ther curatve and preventve effectveness. The curatve effectveness of blanket DCT, or another strategy s usually qute well-known or assumed, compared to ts preventve effectveness, whch remans less predctable. In fact, the magntude of the preventve effect depends wdely on the rsk of new IMI. Currently, n many herds, ths rsk has been wdely reduced by better management and hygene durng the dry perod [5, 17, 33, 71, 72] and consequently, one may only expect a small resultng preventve effect, even wth a very effectve DCT. Relevant knowledge on the rsk of new IMI and on ts preventable fracton by DCT (or applcaton of a teat sealer) s thus crtcal for decson makng. Some revew papers have already been publshed on mastts control durng the dry perod and a large number of orgnal studes on the rsk of new IMI durng the dry perod have been conducted [5, 10, 22, 32, 71]. However, no study has provded a quanttatve overvew of the dfferences n ncdence wth or wthout DCT, nor a dscusson of varaton n these dfferences. Therefore, the purpose of ths paper was to provde a quanttatve comparatve analyss of the ncdence of spontaneously occurrng IMI durng the dry perod, wth or wthout DCT (random or selectve omsson, wth or wthout applcaton of a teat sealer). 2. MATERIALS AND METHODS 2.1. Materals for meta-analyss Selecton of papers A search n the lterature was conducted usng the Commonwealth Abstract Bulletn (CAB) database for papers ndexed snce 1990, Medlne for papers ndexed snce 1966, and OldMedlne for artcles ndexed from 1957 to Research terms conssted of combned descrptors of the dsease, the populaton under study and the perod n queston: (Mastts or mammary gland dsease or ntramammary nfecton or udder

3 Intramammary nfectons durng the dry perod 27 nfecton or ntramammary pathogen or udder pathogen) and (dary cows or dary cattle) and (late lactaton or lactaton end or dry cow or selectve dry cow therapy or systematc dry cow therapy or selectve dry cow treatment or systematc dry cow treatment or antbotherapy or antbotc or dry perod or dryng-off or pre-partum or parturton or post-partum or early lactaton). Only papers publshed n Englsh, French, German, Italan or Spansh were consdered. At ths stage, the references cted n all papers found were also taken nto account. The papers had to fulfl the followng condtons: () To be an orgnal study (revew or symposum papers were cted but not ncluded n the quanttatve analyses). Moreover, f several studes had been made usng the same study sample and method, only the study wth the most detaled descrpton of materals and methods was kept. () To report the ncdence of spontaneous IMI cases durng the dry perod n feld condtons. Routes and nfectve doses of expermentally nduced IMI were not consdered as beng fully representatve of IMI under feld condtons. () The status of quarters regardng IMI had to be determned based on mcrobologcal examnatons of aseptc quarter mlk samples. A total of 36 papers based on 30 studes were ncluded. Some authors presented the results of several protocols n the same paper (for easer readng, each protocol was numbered n roman numerals [5-I, 5-II, 59-I, 59-IV]). Others presented results from the same protocol n several papers (for easer readng, studes on the same protocol were grouped [3 4, 10 11, , 66 74, 76 77, 78 79]). The aspects of study sample and desgn lkely to nfluence external valdty of the results were systematcally checked and recorded by the same abstractor. The descrpton focused on the followng: country, study perod, type of herds (expermental or commercal), number of herds/cows/quarters, method of dryng-off, selecton of herds/ cows, actve-prncple n the DCT used, mlk samplng tmng, mcrobologcal procedure, defnton of an nfected quarter, defnton of a new IMI, prevalence of IMI at dryngoff and type of pathogen nvolved, number of quarters at rsk contractng a new IMI durng the dry perod. All these data are dsplayed n Tables I, II and III, and n Secton Study desgn n selected papers Study sample The study samples n the selected papers ncluded from 1 to 140 herds (manly less than 10 herds) and from 40 to cows (manly around cows) (Tab. I). All studes were done at the quarter level as the observaton unt Type of comparson mplemented Frstly, the effect of a random total omsson of antbotc DCT was assessed. Such studes compared cows from the same herd randomly assgned n two groups: a group of untreated cows and a group of treated cows (type 1a n Fg. 1) [5-I, 5-II, 14, 22, 24, 30, 31, 32, 34, 57, 58, 66 74, 67, 75, 76 77, 78 79, 80, 82, 83]. In order to account for ndvdual factors lkely to nfluence the ncdence of IMI, some authors made an ntra-cow comparson (type 1b n Fg. 1): n each cow, some quarters were treated and some were left untreated. Dependng on the study, quarters were ether randomly assgned [48, 59-I, 88] or a fxed half udder was treated or not [59-IV, 71]. The same study desgns (1a or 1b) were used to compare the applcaton of a teat sealer to DCT [35, 88] or to the absence of any DCT [3 4, 88]. However, these study desgns were appled here to a selected sample of cows or quarters, presumed not nfected or nfected at a low prevalence level (see Tab. I). Secondly, the effect of selectve DCT strateges was assessed. Studes compared cows from the same herd randomly assgned nto two groups: selectvely treated versus

4 28 A. Robert et al. Table I. Study samples n selected studes (sorted by year of publcaton). Reference Country a Number of herds (type)/number of cows b Method of dryng-off c Incluson crtera c [76 77] ZA 35 (C) / 888 na Not stated [59-I] IE 6 (na) / 146 A New nfecton only for unnfected quarters [59-IV] IE 1 (C) / 75 A New nfecton only for unnfected quarters [67] IL 1 (C) / 79 A Not stated [22] US 5 (E) / 222 na Not stated [32] US na(e C) / 964 na Not stated [75] US 1 (E) / 172 na Cows unnfected at dryng-off except by Mcrococc, dphtherods, Staphylococcus epderms [48] US 5 (C) / 252 na Not stated [66 74] FR 1 (E) / 190 na Not stated [31] US 18 (C) / 273 na Not stated [62] US 1 (E) / 232 na Not stated [82] US 75 (C) / A Hgh prevalence of staphylococcal and streptococcal nfectons; C. bovs consdered as not pathogen [24] US 141 (C) / na Not stated [57] NZ 6 (C) / 330 na Not stated [58] NZ 7 (C) / 120 na Not stated [ ] GB 6 (C) / 930 na Herds wth BSCC < c/ml, wth < 15% of quarters nfected by a MP, and free from St. agalactae [78 79] US 1 (E) / 43 A Herd wth < 1% of quarters nfected by CPS and St. agalactae [30] US 1 (E) / 140 na Not stated [14] US 1 (E) / 80 na Not stated [10 11] AU 12 (C) / na Herds wth < BSCC < c/ml; cows nfected n less than 3 quarters; new nfecton only for unnfected quarters [80] IL 1 (C) / 106 na Herd wth BSCC = c/ml; new nfecton only for unnfected quarters [71] NL 1 (E) / 68 I Herd wth BSCC = c/ml, wth < 5% of quarters nfected by a MP [34] US 4 (C) / 185 A Herds wth < BSCC < c/ml; new nfecton only for unnfected quarters [86] NZ 4 (E C) / 371 I New nfecton only for unnfected quarters [88] NZ 3 (C) / 528 na Cows wth ISCC < c/ml, wth at least 3 unnfected quarters at dryng-off [3 4] GB 2 (E) / 170 5(C) / 231 A Herds wth < BSCC < c/ml; cows wth ISCC < c/ml, wth no clncal case [5-I] GB 2 (E) / 236 A Herds wth BSCC = c/ml; cows unnfected at dryng-off except by CNS or Corynebacterum speces [5-II] GB 2 (C) / 76 A Herds wth BSCC = c/ml; cows wth varety of nfectons at dryng-off [35] GB 16 (C) / 467 na Herds wth BSCC < c/ml; cows wth ISCC < c/ml, no clncal case [83] AR 1 (E) / 44 na Cows wth no nfecton wth MP at dryng-off a Internatonal abbrevaton of the country; b (C): commercal; (E): expermental; c A: abrupt, I: ntermttent, na: not avalable; d CNS: coagulase-negatve staphylococc, CPS: coagulase-postve staphylococc, c/ml: cells by mllltre, ISCC: ndvdual somatc cell count, BSCC: bulk tank mlk somatc cell count, MP: major pathogens, C. bovs: Corynebacterum bovs, St. agalactae: Streptococcus agalactae.

5 Intramammary nfectons durng the dry perod 29 Table II. Elements of materals and methods used n selected studes (sorted by year of publcaton). Reference Actve prncple of DCT (n mg/ tube ) a Pathogens ncluded n the calculaton of the ncdence b [76 77] CP 1) Clox (200-LA) / EP 2) Clox (1000-LA) S. au / St. ag, dy, ub [59-I] CP (QT) Pen (300) CPS / St. ub, dys / Other [59-IV] CP (QT) Clox (500-LA) CPS / St. ub, dys / Other [67] CP Clox (500) S. au / St. ag [22] CP- Pen ( I U)+DHS (1000-na) S. au/ St. ag, sp / Cf [32] CP Clox (500) na [75] EP Clox (500-na) S. au / St. sp / Other [48] EP Novo (600)+Pen ( I U-LA) na [66 74] CP Clox (500) / EP Pen (10 6 IU)+DHS (1000) na [31] EP Novo(400)+Pen (10 5, , I U-na) / S. au / St. ag, sp / Other cocc Novo(400, 600-na) / Pen (10 5, , I U-na) [62] CP Pen (10 6 IU)+DHS (1000-LA) S. au / St. ag, S. sp / Cf / Mxed [82] EP Novo (1) 50-LA, 2) 200-LA, 3) 400-LA, 4) S. au / St. ag, sp 600-LA) [24] EP 1) Pen (10 6 IU)+DHS (1000-na) / 2) Clox (500) S. sp / St. ag, sp / Cf / 3) 4) 5) Novo+Pen (du) [57] CP 1) Neo (500) + Pen ( IU-na) / 2) Clox S. au, ep / St. ag, ub, sp (500) [58] CP Pen ( I U) + Novo (400) S. au, ep / St. ag, ub, sp [ ] CP Pen (10 6 IU)+DHS (500-LA) / Clox (500) S. au / St. ub / Cf [78 79] CP Novo (du) St. sp / Cf [30] CP 1) Novo (400-na) / 2) Cephp (300-na) / 3) Pen S. au, CN / St. sp / Cf / C. b / other (10 6 IU)+DHS (1000-na) [14] CP Clox (500) 1) 1 njecton / 2) 3 njectons S. CN, CP / St. sp / C. sp / Ba / Ps / Ye [10 11] CP Clox (500) S. au / St. ub / other [80] CP 1) Norfloxacn (subcutaneous) / 2) S. au Oxytetracycln (ntramuscularly) / 3) Cephp (500-na) [71] CP (QT) Pen (10 6 IU)+DHS (100) + nafclln (100-LA) S. au, CN / St. ub, dy, sp / Cf / A. p / Ent / Ps [34] CP Cephp (300) S. au, sp / St. sp / Cf / C. b [86] CP Cephl (250-LA) S. au, CN/ St. ub / C. b [88] CP (QT) Teat Sealer / Cephl (250-LA) / Teat sealer S. au, CN/ St. sp. / Cf / other + Clox (600) [3 4] CP Teat Sealer S. au / St. ub, dy / Cf / A. sp / Mxed [5] CP Cephl (250-LA) / Clox (600-LA) S. au / St. ub, dy / Cf / A. sp / Mxed [35] CP Teat Sealer / Cephl (250-LA) S. au, CN / St. ub, dy, sp / Cf / A. p / Ent / Ps [83] NA Spramycn + DHS (ntramuscularly) S.au / St. ub, dy a CP: commercal preparaton; EP: expermental preparaton; NA: nformaton about type of preparaton not avalable; Cephl: cephalonum; Clox: cloxaclln; Neo: neomycn; Pen: penclln; Novo: novobocn; DHS: dhydrostreptomycn; Cephp: cephaprn; (du: dose unknown; quantty of actve prncple n mg LA: long actng ontment base; na: base ontment not avalable; when nothng s specfed t means short actng base ontment); b S.: Staphylococcus (au: aureus, CN: coagulase negatve, CP: coagulase postve, ep: epderms, sp: speces); St.: Streptococcus (ag: agalactae, (ag): S. ag excluded, dy: dysgalactae, gr: green (ubers, dysgalactae, group D), sp: speces, ub: ubers); C: Corynebacterum (b: bovs, sp: speces); Ye: Yeast; Cf: Colform ; Ent: Enterobacteraceae (E. c: Eschercha col), Ba: Bacllus speces, Ps: Pseudomonas sp; A. p: Arcanobacterum pyogenes; na: dentfcaton not made avalable.

6 30 A. Robert et al. Table III. Quarter ncdence of ntramammary nfectons durng the dry-perod by pathogen type n quarters treated and untreated at dryng-off. Incdence of ntramammary nfecton c Reference a No. b All Streptococc d Staphylococc e Colform Others Treated Untreated Treated Untreated Treated Untreated Treated Untreated Treated Untreated Type 1a [76 77] *** *** au 1.6 au 5.3*** *** ** au 0.3 au 5.3*** [67] # ag 1.1 ag 4.5 NS au 5.5 au 27.2*** [22] # ** ag 0.4 sp 2.2 ag 1.4 NS sp 3.4 NS au 0.2 au 2.5** NS NS [32] *** [75] *** [66 74] # NS [31] NSA ag 0.4 sp 1.4 ag 2.0 sp 3.0 au 2.6 au [82] # NS ag 1.4 sp 3.5 ag 2.1 NS sp 5.5* au 4.4 au 2.6* *** ag 0.6 sp 2.2 ag 2.1** sp 5.5*** au 3.2 au 2.6 NS * ag 0.4 sp 3.7 ag 2.1** sp 5.5 NS au 2.9 au 2.6 NS *** ag 0.4 sp 2.9 ag 2.1** sp 5.5** au 2.7 au 2.6 NS [24] NSA ag 0.8 sp 1.9 ag 1.5 sp ag 0.0 sp 3.5 ag 1.5 sp ag 0.4 sp 2.1 ag 1.5 sp ag 0.0 sp 2.8 ag 1.5 sp ag 0.4 sp 1.9 ag 1.5 sp [57] # ** ag 0.5 ub 3.3 ag 2.2* ub 4.5 NS au 4.0 au 7.9* NS ag 0.0 ub 3.6 ag 2.2 NS ub 4.5 NS au 5.4 au 7.9 NS [58] # ub 3.6 ub 3.9 NS au 3.6 au 7.8 NS [78 79] # NS NS [30] NS NS au 0.0 CN 6.3 au 0.8 NS CN 6.9 NS NS NS ** NS au 0.0 CN 1.3 au 0.8 NS CN 6.9* NS NS NS NS au 0.0 CN 12.9 au 0.8 NS CN 6.9 NS NS NS

7 Intramammary nfectons durng the dry perod 31 Table III. Contnued. Incdence of ntramammary nfecton c Reference a No. b All Streptococc d Staphylococc e Colform Others Treated Untreated Treated Untreated Treated Untreated Treated Untreated Treated Untreated [14] # NS NS CP 2.0 CN 7.1 CP 7.6 NS CN 4.3 NS NS NS * CP 1.0 CN 10.3 CP 7.6 NS CN 4.3 NS NS [80] 1 CP 17.1 CP 29.3* 2 CP 9.7 CP 29.3** 3 CP 51.7 CP 29.3* [34] * * au 1.1 sp 12.2 au 2.1 NS sp 13.5 NS NS [5-I] *** dy 0.0 ub 1.4 dy 0.6 NS ub 6.5*** au 0.2 au 1.4 NS NS NS [5-II] *** dy 0.0 ub 0.0 dy 2.9 NS ub 9.3*** au 0.0 au 7.0** NS NS [83] # * dy 2.3 ub 0.0 dy 1.1 NS ub 8.0* aur 0.0 au 2.3 NS Type 1b [59-I] QT NS sp 1.4 sp 3.0 NS CP 11.0 CP 11.8 NS NS [59-IV] ½T NS sp 2.8 sp 4.7 NS CP 7.4 CP 5.7 NS NS [48] QT # ** [71] ½T # NS au 0.0 CN14.9 au 1.1 NS CN 6.4 NS NS [88] QT *** sp 1.0 sp 9.5*** CP 0.0 CN 0.2 CP 0.6 NS CN 1.1 NS NS NS a QT: per cows, quarters were randomly allocated to the treated/untreated group; ½T: the same half udder was assgned to group of quarters treated/untreated; #: statstcal analyss done by revewers; : ncdence of ntramammary nfectons durng the dry perod calculated by revewers; NSA: no statstcal analyss could have been done; Type 1a: antbotc treatment was randomly appled at the cow level; Type 1b: antbotc treatment was randomly appled at the quarter level. b Number of the dry cow treatment used (see Tab. II). c Incdence of ntramammary nfectons durng the dry perod (n percentage of quarters): number of quarters newly nfected durng the dry perod dvded by the total number of quarters at rsk of new ntramammary nfectons durng the dry perod. d ag: Streptococcus agalactae, dy: Streptococcus dysgalactae, sp: Streptococcus speces, ub: Streptococcus ubers. e au: Staphylococcus aureus, CN: coagulase-negatve staphylococc, CP: coagulase-postve staphylococc, sp: Staphylococcus speces. * Sp < 0.05; ** Sp < 0.01; *** Sp < 0.001; NS: not sgnfcant.

8 32 A. Robert et al. Type 1 a Type 1 b Type 2 b Type 2 a Fgure 1. Type of comparson n selected studes.

9 Intramammary nfectons durng the dry perod 33 blanket DCT. The selectve treatment could be appled at two levels: at the cow level (selectve DCT) for type 2a n Fgure 1 [10 11, 62, 66 74, 86] and at the quarter level (selectve DQT) for type 2b n Fgure 1 [10 11, , 86]. The selecton crtera of cows or quarters treated were based on mcrobologcal examnaton [10 11, , 86] or on the SCC level and/or the Calfornan mastts test score [62, 66 74] Mlk samplng for bacterologcal examnaton In all selected studes, samples were taken at dryng-off and shortly after calvng, to assess the possble change n quarter nfecton status durng the dry perod. In 24 studes, two (or more) consecutve samples (at a few days nterval) [3 4, 5-I, 5-II, 10 11, 22, 62, 66 74, 75, 78 79] or two (or more) duplcate samples (at the same tme) [14, 24, 31, 32, 34, 35, 57, 58, 67, 71, 76 77, 82, 83, 86, 88] were taken. In one study, some addtonal samples were taken along the dry perod, to better assess the tmng of the occurrence of new IMI [78 79] Bacterologcal procedure The two standard procedures mostly used were nsttuted by the Natonal Mastts Councl [14, 22, 24, 30, 57, 58, 62, 78 79, 82] or by the Internatonal Dary Federaton [3 4, 5-I, 5-II]. Some other authors used a personal procedure derved from the prevous ones. Methodologcal ssues whch could mpar the senstvty or specfcty of the bacterologcal procedures were checked: the volume of mlk spread over a fxed area [12], the storage of samples, especally the freezng of samples [39, 70], and the type of growth medum used (selectve medum or not) Antbotc treatment Ths paper was not amed at summarsng the curatve effectveness of DCT. The pharmacoknetc propertes of the ontment base were therefore not consdered. Most DCT were ntramammary-labelled drugs (for detals see Tab. II), except n [80] and [83], where some DCT were admnstered ntramuscularly or subcutaneously. The molecules used were specfed only to explan possble dscrepances among the studes. In the selected studes, some molecules wth a wde spectrum of actvty on Gram+ and Gram pathogens [5-I, 22, 24, 30, 34, 35, 48, 57, 58, 59-I, 62, , 66 74, 71, 80, 83, 86, 88], and also other molecules only actve on Gram+ pathogens [5-II, 10 11, 14, 24, 30, 31, 32, 57, 59-IV, , 67, 66 74, 75, 76 77, 78 79, 82, 88] were used Defnton of a new ntramammary nfecton A quarter was consdered nfected f the same bacteral speces was solated n two (or more) consecutve or duplcate samples. A new IMI occurrng durng the dry perod was declared when a pathogen, not found at dryng-off, was found at calvng. Ths defnton ncludes two dfferent quarter nfecton statuses at dryng-off: () unnfected at dryng-off and nfected at calvng and () nfected by a pathogen at dryng-off and nfected by another pathogen at calvng. Some authors dd not consder the second defnton. They ether dd not observe such a new IMI (due to ts very low frequency of occurrence) or they excluded all quarters already nfected at dryng-off from those at rsk of a new IMI (most common) Calculaton of the ncdence of ntramammary nfectons The ncdence of IMI was the number of quarters contractng a new IMI durng the dry perod dvded by the total number of quarters at rsk of new IMI at dryng-off. Accordng to the defnton of a new IMI (gven above), the total number of quarters at rsk could be dfferent: all quarters versus

10 34 A. Robert et al. only quarters unnfected at dryng off [10 11, 34, 59-I, 59-IV, 80, 86]. In some papers, the ncdence was drectly provded. For others [14, 22, 48, 71, 83], calculatons were made by revewers. In each study of type 1a and 1b (Fg. 1), the ncdence n untreated quarters was compared to the ncdence n treated quarters. In each study of type 2a and 2b (Fg. 1), the ncdence n cows under selectve DCT or DQT was compared to the ncdence n cows under blanket DCT. The sgnfcance of the dfference n ncdence between groups was assessed by a χ 2 test, when t was feasble and not already provded by the authors of the orgnal papers [14, 22, 48, 57, 58, 66 74, 67, 71, 78 79, 82, 83] Meta-analyss procedure The relatve rsk of new IMI (the ncdence n untreated quarters versus the ncdence n treated quarters) was summarsed by a meta-analyss. The general varancebased method descrbed by [60] was used. A summary estmate of the relatve rsk (RR s ) was determned for all pathogens and also by pathogen type. The formulae for meta-analyss were the followng: ln RR w = s ( w ln RR ) = CI w 1 var IUT RR = IT 1 RR± 1.96 w = e ( UT + T ) NNI var = UT T NI where RR s s the summary estmate of the relatve rsk of new IMI (ncdence n untreated quarters versus ncdence n treated quarters), w s the weght assgned to the th study, RR s the relatve rsk of new IMI (ncdence n untreated quarters versus ncdence n treated quarters) n the th study, CI are the lmts of the 95% confdence nterval of the summary estmate of the relatve rsk of new IMI, var s the varance of the relatve rsk of new IMI n the th study, UT s the total number of quarters at rsk among untreated quarters n the th study, T s the total number of quarters at rsk among treated quarters n the th study, NNI s the total number of quarters not contractng a new IMI among both treated and untreated quarters n the th study, NI s the total number of quarters contractng a new IMI among both treated and untreated quarters n the th study, IUT s the ncdence n untreated quarters (n % of quarters at rsk) n the th study, IT s the ncdence n treated quarters (n % of quarters at rsk) n the th study. In order to assess the relevance of the summary estmates of the relatve rsk, the Q parameter was calculated and compared to a χ 2 dstrbuton, wth a number of degrees of freedom equal to the number of studes mnus 1 [60]. Ths test of homogenety checks f all studed samples can be assumed to come from the same meta-populaton. Q was calculated as: 2 ( w ( RR RR ) ) Q = In case of heterogenety, as an attempt to partally explan the devaton, a metaregresson was done [29]. The model was: Q = β + β YEAR + β BACT + β DCT β 4 PREV + β 5 Staph + ε where Q s the weghted dfference between the summary estmate of the relatve rsk and the relatve rsk n the th study, β 0 s the ntercept, β 1 YEAR s the year of publcaton n the th study (after versus before 1985), β 2 BACT s the assessment of the mcrobologcal procedures n the th study (one sample versus more than 1), β 3 DCT s the DCT used n the th study (Gram+ alone versus Gram+ and Gram actvty spectrum), β 4 PREV s the prevalence at dryngoff n the th study (less versus more than 25% of nfected quarters), β 5 Staph s the s.

11 Intramammary nfectons durng the dry perod 35 Fgure 2. Incdence n untreated cows durng the dry perod (n % of quarters at rsk) and prevalence at dryng-off n untreated cows (n % of quarters at rsk). proporton of quarters nfected by staphylococc (CPS and CNS) at dryng-off n the th study (less versus more than 15% of nfected quarters), ε s the resdual error term. Cut-off values were selected accordng to the dstrbuton of the varables. If a varable was sgnfcantly assocated wth the Q (P value < 0.20), a secondary meta-analyss was mplemented, after stratfcaton of the studes accordng to the levels of ths varable. To calculate weghted means and meta-analytc RR s, studes could only be consdered once when several DCT were used n the same study. Therefore, two summary estmates of the relatve rsk were provded. To obtan the frst one, DCT whch mnmsed the dfference between untreated and treated quarters, and for the second one, DCT whch maxmsed ths dfference, were retaned. Nether meta-analyss nor weghted mean calculatons were deemed relevant when less than four studes could be ncluded. 3. RESULTS 3.1. Incdence followng random omsson of antbotc treatment Comparson between untreated and antbotc-treated quarters (type 1a and 1b desgns) Cumulated ncdence of ntramammary nfectons In untreated quarters at rsk, 6.5 to 33.3% contracted a new IMI n studes of type 1a, and 11.6 to 35.1% n studes of type 1b (Tab. III), and the ncdence averaged 12.8% (weghted mean). The ncdence was not correlated to the prevalence of IMI n untreated quarters at dryng-off (Spearman correlaton coeffcent = 0.25, P value = 0.33) (Fg. 2). In antbotc-treated quarters at rsk, 0 to 28.6% contracted a new IMI n studes of type 1a and 2.3 to 37.2% n studes of type 1b

12 36 A. Robert et al. Fgure 3. Incdence n untreated cows durng the dry perod (n % of quarters at rsk) and dfference n ncdence between untreated and treated cows (untreated mnus treated cows) (n % ponts). (A) For all pathogens, (B) for streptococc, (C) for coagulase-postve staphylococc.

13 Intramammary nfectons durng the dry perod 37 (Tab. III), and the ncdence averaged 6.6 or 8.0% (weghted means), accordng to the DCT used (respectvely, the most and the less effectve). A hgher ncdence was systematcally observed n untreated than n treated quarters, except n [71] (Fg. 3). The dfference (ncdence n untreated quarters mnus ncdence n treated quarters) vared from 2.1 to +22.1% ponts (Fg. 3), beng sgnfcant n 15 studes out of 25 (Tab. III). Intrastudy, the relatve rsk ranged from 0.94 to 5.69 (Tab. IV) and the summary RR s were 1.54 and 1.94 (Tab. V), dependng on the DCT (respectvely, the most and the less effectve). The dfferences were heterogeneous between studes. None of the factors tested by the meta-regresson establshed a possble orgn for the heterogenety of the studes. No secondary meta-analyss was done Incdence by pathogens New IMI were manly due to streptococc and CPS. In 50% of the studes, each of these pathogens represented more than 35% of new IMI (Tab. III) Streptococc In untreated quarters at rsk, 4.0 to 12.5% contracted a new IMI due to streptococc n studes of type 1a and 3.0 to 9.5% n studes of type 1b (Tab. III), and the ncdence averaged 5.9% (weghted mean). In antbotc-treated quarters at rsk, 0.0 to 10.2% contracted a new IMI n studes of type 1a and 1.0 to 2.8% n studes of type 1b (Tab. III), and the ncdence averaged 1.9 or 2.7% (weghted means), accordng to the DCT used (respectvely, the most and the less effectve). A hgher ncdence was systematcally observed n untreated than n treated quarters n all studes, except n [14, 83] (Fg. 3). The dfference (ncdence n untreated quarters mnus ncdence n treated quarters) vared from 2.8 to +8.5% ponts (Fg. 3), beng sgnfcant n half of the studes (Tab. III). Intra-study, the relatve rsk ranged from 0.75 to (Tab. IV) and the estmates of RR s were 2.32 and 3.15 (Tab. V), dependng on the DCT used. None of the factors tested by the meta-regresson establshed a possble orgn for the heterogenety between the studes Coagulase-postve staphylococc In untreated quarters at rsk, 0.8 to 29.3% contracted a new IMI due to CPS n studes of type 1a and 0.6 to 11.8% n studes of type 1b (Tab. III), and the ncdence averaged 4.8% (weghted mean). In antbotc-treated quarters at rsk, 0.0 to 51.7% contracted a new IMI n studes of type 1a and 0.0 to 11.0% n studes of type 1b (Tab. III), and the ncdence averaged 1.9 to 3.0% (weghted means), accordng to the DCT used. A hgher ncdence was observed n untreated than n treated quarters n the majorty of the studes (15 among 18) (Fg. 3). The dfference (ncdence n untreated quarters mnus ncdence n treated quarters) vared from 22.4 to +21.7% ponts (Fg. 3), beng sgnfcant n only fve studes (Tab. III). Intra-study, the relatve rsk ranged from 0.56 to (Tab. IV) and the RR s were 1.34 and 3.04 (Tab. V), dependng on the DCT used. None of the factors tested by the meta-regresson establshed a possble orgn for the heterogenety between the studes Coagulase-negatve staphylococc Only fve studes [14, 30, 34, 71, 88] reported the specfc ncdence of CNS IMI (Tab. III). In untreated quarters at rsk, 4.3 to 13.5% contracted a new IMI n studes of type 1a and 1.1 to 6.4% n studes of type 1b (Tab. III), and the ncdence averaged 5.6% (weghted mean). In antbotc-treated quarters at rsk, 1.3 to 12.9% contracted a new IMI n studes of type 1a and 0.2 to 14.9% n studes of

14 38 A. Robert et al. Table IV. Relatve rsk of new ntramammary nfectons durng the dry-perod by pathogen type (ncdence n untreated quarters versus ncdence n treated quarters). Relatve rsk of new ntramammary nfectons c Reference a No. b All Streptococc Coagulase postve Coagulase negatve pathogens staphylococc staphylococc Type 1a Colform [76 77] [67] # 4.94 [22] # [32] [75] [66 74] # [31] NSA [82] # [24] NSA [57] # [58] # 2.16 [78 79]# [30] [14] # [80] [34] [5-I] [5-II] [83] # Type 1b [59-I] QT [59-IV] ½T [48] QT # 1.46 [71] ½T # [88] QT a QT: per cows, quarters were randomly allocated to treated/untreated group; ½T: the same half udder was assgned to a group of treated/untreated quarters; #: statstcal analyss done by revewers; : ncdence of new ntramammary nfectons durng the dry perod calculated by revewers; NSA: no statstcal analyss could have been done; Type 1a: antbotc treatment was randomly appled at the cow level; Type 1b: antbotc treatment was randomly appled at the quarter level. b Number of the dry cow treatment used (see Tab. II). c Relatve rsk of new ntramammary nfectons durng the dry perod: ncdence of ntramammary nfectons durng the dry perod n untreated quarters versus ncdence n treated quarters.

15 Intramammary nfectons durng the dry perod 39 Table V. Summary estmate, confdence nterval and test for homogenety of the relatve rsk of new ntramammary nfectons durng the dry perod (ncdence n untreated quarters versus ncdence n treated quarters) from meta-analyss. Pathogen type Reference Mnmal RR of new ntramammary nfectons Maxmal RR of new ntramammary nfectons RR s a CI b Q c RR s a CI b Q c All pathogens [5-II, 14, 22, 24, , 34, 48, 57, 59-I, 59-IV, 71, 66 74, 76 77, 82, 83, 88] Streptococc Coagulase postve staphylococc 1.54 [1.42;1.67] (s) 1.94 [1.78;2.13] (s) [14, 22, 24, 30, 31, 34, 59-I, 2.32 [1.98;2.72] (s) 3.15 [2.63;3.78] (s) 59-IV, 76 77, 78 79, 82, 88] [5-II, 14, 22, 31, 34, 57, 58, 1.34 [1.10;1.64] (s) 3.04 [2.42;3.81] (s) 59-I, 59-IV, 67, 76 77, 80, 82] Coagulase negatve staphylococc [14, 30, 34, 71, 88] 0.83 [0.61;1.14] (s) 1.12 [0.80;1.57] (s) Colform [5 I, 22, 24, 30, 78 79, 88] 0.86 [0.62;1.20] 1.55 (ns) 0.96 [0.70;1.34] 2.75 (ns) a RR s : summary estmate of the relatve rsk of new ntramammary nfectons durng the dry perod (ncdence n untreated quarters versus ncdence n treated quarters). b CI: 95% confdence nterval of the summary estmate of the relatve rsk of new ntramammary nfectons durng the dry perod. c Q: test for homogenety, (s): heterogeneous, (ns): homogeneous. type 1b (Tab. III), and the ncdence averaged 5.1 or 6.7% (weghted means), accordng to the DCT used. Three studes reported a hgher ncdence of CNS IMI n untreated than n treated quarters and two reported a lower ncdence. The dfference (ncdence n untreated quarters mnus ncdence n treated quarters) vared from 8.5 to +5.5% ponts, beng sgnfcant n one of the studes (Tab. III). Intrastudy, the relatve rsk ranged from 0.42 to 6.00 (Tab. IV) and the RR s were 0.83 and 1.12 (Tab. V), dependng on the DCT used. None of the factors tested by the metaregresson establshed a possble orgn for the heterogenety between the studes Colforms A small number of studes (8 studes) reported new Colform IMI. In the others, the authors ether dd not observe Colform IMI or only focused on Gram+ pathogens. In untreated quarters at rsk, 0.0 to 5.6% contracted a new IMI n studes of type 1a and 0.8% n studes of type 1b (Tab. III), and the ncdence averaged 1.2% (weghted mean). In antbotc-treated quarters at rsk, 0.0 to 11.0% contracted a new IMI n studes of type 1a and 0.8% n studes of type 1b (Tab. III), and the ncdence averaged 1.2 to 1.3% (weghted means), accordng to the DCT used. No clear trend was observed for the dfference n ncdence between the untreated and the treated quarters, whch vared from 2.8 to +8.5% ponts, beng sgnfcant n half of the studes (Tab. III). Intra-study, the relatve rsk ranged from 0.50 to 1.49 (Tab. IV) and the RR s were 0.86 and 0.96 (Tab. V), dependng on the DCT used. Dfferences were homogeneous between the studes.

16 40 A. Robert et al. Table VI. Incdence of ntramammary nfectons durng the dry perod and relatve rsk of new ntramammary nfectons durng the dry perod n groups of quarters untreated, antbotc treated, or wth a teat sealer applcaton. Incdence of ntramammary nfectons b Relatve rsk of new ntramammary nfectons c Reference a Teat sealer Antbotc Untreated Teat sealer versus Teat sealer versus treated antbotc treated untreated [88] QT # ( 0.1 NS) 12.7 (+10.3***) [35] # (+6.4**) 0.85 [3 4] (+8.2***) 0.29 a QT: per cows, quarters were randomly allocated to the antbotc treated group, the untreated group or the group wth a teat sealer applcaton; #: statstcal analyss done by revewers; : ncdence of new ntramammary nfectons durng the dry perod calculated by revewers. b Incdence of ntramammary nfectons durng the dry perod (n percentage of quarters): number of quarters newly nfected durng the dry perod dvded by the total number of quarters at rsk of new ntramammary nfectons durng the dry perod, n the groups wth a teat sealer applcaton, the antbotc treated group, or the untreated group. c Relatve rsk of new ntramammary nfectons durng the dry perod: rato of ncdence n quarters wth a teat sealer applcaton versus antbotc treated or untreated quarters. * Sp < 0.05; ** Sp < 0.01; *** Sp < 0.001; NS: not sgnfcant Comparson of applcaton of a teat sealer to ether the absence of antbotc treatment or blanket dry cow treatment No study on external teat sealer was elgble for ths meta-analyss and only 3 elgble studes [3 4, 35, 88] dealt wth an nternal teat sealer made of bsmuth subntrate n an oly bass. Study samples were restrcted to cows, or quarters presumed not nfected or nfected at a low prevalence level (see Tab. I). In the selected studes, the ncdence n the untreated quarters was qute smlar to the studes of types 1a and 1b presented above, and bd. n the treated quarters. When a teat sealer was appled, from 2.4 to 36.2% of the quarters at rsk contracted a new IMI (Tab. VI). New IMI were manly due to CNS (above 50% of new IMI). The ncdence was sgnfcantly hgher n untreated quarters compared to quarters wth an applcaton of a teat sealer. Moreover, the ncdence was lower (sgnfcantly n [35], and not sgnfcantly n [88]) n quarters wth applcaton of a teat sealer than n antbotc-treated quarters (Tab. VI). Intra-study, the relatve rsks were 0.85 or 1.04 and 0.19 or 0.29, for quarters wth an applcaton of a teat sealer versus, respectvely, antbotctreated and untreated quarters (Tab. VI). Gven the small number of studes, no weghted mean, nor RR s were calculated Incdence followng selectve omsson of antbotc treatment In these studes, the DCT was not randomly appled to groups derved from the overall populaton, but only to a sub-populaton of selected cows or quarters (Fg. 1). The selecton crtera used (Tab. VII) ndcate that the cows left untreated were probably unnfected or only nfected wth a low probablty. In the group of cows under blanket DCT, from 2.6 to 21.9% of quarters at rsk contracted a new IMI and the average ncdence was 6.5% (weghted mean). Ths level was not dscrepant wth the results of studes prevously analysed (Sect. 2.1). In cows under selectve DCT, from 3.9 to 28.7% of the quarters at rsk contracted a new IMI, and the average ncdence was 9.9% (weghted mean), when 27.1 to 57.1%

17 Intramammary nfectons durng the dry perod 41 Table VII. Incdence of new ntramammary nfectons durng the dry perod dependng on antbotc treatment strategy at dryng-off: blanket dry cow treatment versus selectve dry cow treatment or dry quarter treatment. Reference a Prevalence at dryng-off Incdence (n % of quarters) e % treated c Crteron of treatment d Relatve Blanket DCT Selectve DCT Selectve DQT rsk f At cow level [66 74] # CMT score > +++ (DO-30 days) (+6.8% NS) 1.31 [62] ISCC > cell/ml (DO-30 days) or CMT score > ++ n any quarter at DO or past hstory of clncal mastts (+3.4% *) 2.07 [10 11] # one or more quarter (s) nfected (+1.3% NS) 1.50 [86] # one or more quarter (s) nfected (+4.3% ***) 2.23 At quarter level [ ] nfected by a major pathogen at DO (+7.0% *) 1.83 [10 11] # only nfected quarter (s) (+ 3.8% ***) 2.45 [86 ] # only nfected quarter (s) (+ 5.7% ***) 2.64 a #: statstcal analyss done by revewers; : ncdence of new ntramammary nfectons durng the dry perod calculated by revewers. b Prevalence of nfected quarters at dryng-off: number of quarters nfected at dryng-off dvded by the total number of quarters. c Numbers of treated quarters n selectve DCT or DQT (n percentage of quarters): number of quarters treated dvded by the total number of quarters n the treatment group. d CMT: Calforna mastts test; DO: dryng-off; ISCC: ndvdual somatc cell count. e Incdence of ntramammary nfectons durng the dry perod: number of quarters newly nfected dvded by the total number of quarters at rsk of new ntramammary nfectons durng the dry perod; n brackets the dfference of ncdence of ntramammary nfectons durng the dry perod and the level of sgnfcance between selectve DCT or DQT and blanket DCT; Selectve DCT: selectve dry cow antbotc treatment; Selectve DQT: selectve dry quarter antbotc treatment; Blanket DCT: blanket dry cow antbotc treatment. f Relatve rsk of new ntramammary nfectons durng the dry perod dependng on level = ncdence of new ntramammary nfectons durng the dry perod n selectve DCT versus n blanket DCT or ncdence of new ntramammary nfectons durng the dry perod n selectve DQT versus n blanket DCT. * Sp < 0.05; ** Sp < 0.01; *** Sp < 0.001; NS: not sgnfcant.

18 42 A. Robert et al. of the quarters were treated (Tab. VII). The dfference n ncdence (the ncdence n cows under selectve DCT mnus the ncdence n cows under blanket DCT) ranged from +1.3 to +6.8% ponts, and was sgnfcant n half of the studes. Intra-study, the relatve rsk ranged from 1.31 to 2.23 (Tab. VII) and the summary RR s (selectve DCT versus blanket DCT) was 1.71 (CI = [1.37; 2.14], and Q = 3.49,.e. absence of heterogenety). In quarters under selectve DQT, from 6.4 to 15.5% of the quarters at rsk contracted a new IMI (Tab. VII), when only around 10% of the quarters were treated. The dfference n ncdence (ncdence n cows under selectve DQT mnus ncdence n cows under blanket DCT) ranged from +3.8 to +7.0% (Tab. VII). All studes demonstrated a sgnfcant dfference. Nether weghted mean, nor RR s were calculated from the three selected studes, but the ntra-study relatve rsks vared from 1.83 to 2.64 (Tab. VII). When compared to selectve DCT, the selectve DQT was found assocated wth a hgher ncdence of IMI (RR = 1.64 n [9, 10] and 1.18 NS n [86]). 4. DISCUSSION Ths study was amed at summarsng the comparatve ncdence of spontaneously occurrng IMI durng the dry perod, wthout or wth DCT (random or selectve omsson, wthout or wth applcaton of a teat sealer). New IMI were manly due to streptococc and CPS. A hgher ncdence of streptococcal and CPS IMI was observed n untreated than n treated quarters. No clear trend was observed for the dfference n ncdence between untreated and treated quarters for CNS or Colform IMI. The same result was for the dfference n ncdence observed between selectve DCT and blanket DCT. The ncdence of IMI under selectve DQT was hgher than under blanket DCT. The ncdence of IMI after the applcaton of a teat sealer was lower than n untreated quarters and not dfferent than treated quarters. However, wde varaton was present between the 30 studes (results from 1966 to 2003 n 10 dfferent countres) reported n the 36 selected papers here consdered for revew and meta-analytc calculaton. Only studes based on a comparatve study desgn ncludng an untreated group were ncluded n the analyss, and therefore, a qute large number of papers comparng alternatve DCT wthout any untreated group were not consdered here. Both study populatons and study desgns dffered between the selected papers. Consequently, the ncdence of IMI n untreated quarters vared wdely (from 6.5 to 35.1%), as well as the type of pathogens responsble for these IMI. Ths may possbly explan that, assumng the exstence of addtonal dfferences n rsk factors and n the type of DCT, the test for homogenety of dfferences n relatve rsks between studes concluded to heterogenety. The meta-regresson procedure mplemented here was ntended to dentfy some factors of heterogenety, but remaned unsuccessful. Ths means that sample characterstcs and materal descrptons n studes were nsuffcent to cluster studes or that other varaton factors of ncdence of IMI were probably nvolved. Nevertheless, we provded a summary estmate of the relatve rsk n order to have an approxmate value for the central trend, when the number of studes was not deemed too small. Of course, such summary estmates should only be extrapolated wth cauton to a partcular populaton. In the last decades, the dstrbuton of streptococc has changed consderably. In the older studes, Streptococcus agalactae was frequently observed, whle n the more recent studes, Streptococcus agalactae was less frequent or absent [7, 69]. However, the effect of ths evoluton on the varaton of ncdence of IMI could not be assessed, because of the low number of new IMI due to Streptococcus agalactae. Some methodologcal orgns for possble dfferences between studes were reported n Secton 2.1.2, although the selected papers

19 Intramammary nfectons durng the dry perod 43 dd not all report the same characterstcs. For example, the mcrobologcal procedure (especally duplcate sample, volume of mlk spread, storage of samples, culture medum used, etc.) was dfferent among studes. However, the resultng dfferences n senstvty (ablty to avod false negatves) can be assumed to affect to a larger extent the measurement of the ncdence, but ths effect was neutralsed when calculatng the ntra-study relatve rsk for untreated versus treated quarters, and thus the summary estmate of relatve rsk provded by our calculatons. Many manageral rsk factors of new IMI were evdenced n prevous studes: method of dryng-off, housng condton, teat dppng at dryng-off, vtamn and mneral supplement, (for revew, see [17, 18, 21, 89]). However, when thnkng to omsson or substtutes to DCT, the fact that the technque of ntramammary admnstraton s suspected to be a rsky procedure should be taken nto account. The ncdence of IMI durng the dry perod was found hgher n cows wth a placebo than n untreated or treated cows: 12.6% versus 5.9 n untreated cows or 9.9% n treated cows [9], and 19.4% versus 14.4% n treated cows [16]. Around 60% of new IMI were due to staphylococc n [16], whch are naturally present n the mcroflora of the teat skn [15]. Nocarda asterodes mastts cases were reported after DCT [23, 81], but no lnk wth DCT could be clearly made. The nserton of a pathogen can also occur wth an nternal teat sealer [3 4, 35, 88]. No or lttle nformaton about herd management and qualty n mplementaton of ntramammary DCT admnstraton were avalable n the papers selected and, therefore, no further analyss on ther mpact could be done here. Under feld condtons of DCT (or teat sealer) use, less precautons than those taken n the ncluded studes are mplemented (no careful dsnfecton for aseptcally samplng before nfuson), and then the rsk of new IMI could be hgher than that reported here. To summarse, the central trend for the overall effectveness of DCT was estmated at 33 to 53% relatve reducton of the ncdence of IMI at the quarter level durng the dry perod. The effectveness was found qute dfferent accordng to pathogens (Tab. III). The hghest value (57 to 68%) was reached for streptococcal IMI, and a qute nterestng one (23 to 67%) exsted for CPS IMI. Protecton of DCT aganst CNS and Colforms seemed of lmted magntude or not effectve, n terms of a central trend, gven our analyss. There are two perods of elevated rsk of new IMI durng the dry perod: the frst weeks after dryng-off durng nvoluton of the udder [50] and the weeks precedng calvng durng colostrogeness [13, 53, 85]. Most of the DCT used dd no longer prevent new IMI durng the latter perod of elevated rsk, especally for long dry perods [5, 58, 62]. As a result, at the end of the dry perod, untreated and treated cows would stay at the same rsk of new IMI. All DCT used n the selected studes were effectve aganst Gram+ bactera, but most of them were a pror not effectve aganst Gram bactera. Thus, no clear concluson may be drawn for these latter bactera. By contrast, the lmted or nonexstent effect aganst CNS IMI suggests that these IMI occur very late n the dry perod, just before calvng as well as n early lactaton [43, 52, 54, for revew see 55] or another hypothess s that the resstant CNS stran possbly nfected the udder [40]. From the selected studes, the applcaton of an nternal teat sealer was able to prevent new IMI, qute possbly better than DCT. However, ths result should only be extrapolated wth cauton, due to the small number of elgble comparatve approaches, whch were conducted n a low rsk context, except [35]. In ths paper, selectve DCT, appled n a subpopulaton of lkely unnfected or few nfected cows, was found assocated wth a varable level of elevated ncdence of IMI durng the dry perod, compared to blanket DCT. In counterpart, the proporton of