Antiretroviral therapy was administered to 121 of 235 women (515

Antiretroviral therapy was administered to 121 of 235 women (51.5%) prior to pregnancy; there was an increase from four of 37 (10.8%) in 1994–1999 to 117 of 204 (57.4%) in 2000–2008. ART was initiated within the first 14 weeks of pregnancy in 42 of 235 women (17.9%), ranging from three of 49 (6.1%) in 1994–1999 to 39 of 206 (18.9%) in 2000–2008. Seven per cent started treatment during the third trimester and 19 women (7.9%) never received ART, mainly because of unknown HIV

status during pregnancy and labour (Table 1). From year 2000, only seven women (3.4%) Navitoclax research buy did not receive ART in pregnancy; three women because of late diagnosis and two women because they refused to take any medication. In two cases the reason for no treatment was not given. Information about timing of ART initiation was not available for 20 women (7.8%). Changes in the use of ART over time are shown in Figure 1. ZDV monotherapy was used

most frequently for the prevention of MTCT in the mid 1990s. From 1997 dual drug therapy was used in some women, in 1998 HAART was introduced and from 2000 nearly all women received HAART, often in a regime including a protease inhibitor Talazoparib (PI). Because of concerns about teratogenic side effects of efavirenz, most of the 14 women in our study receiving this drug had it replaced with a PI early in pregnancy. The most commonly used HAART regimens were ZDV+lamivudine (3TC)+lopinavir+ritonavir (62 of 205; 30.2%) and ZDV+3TC+nelfinavir (47 of 205; 22.9%). Nearly 80% of the women continued their ART after delivery. Pneumocystis RAS p21 protein activator 1 jiroveci prophylaxis was administered to 17 of 203 pregnant women (8.4%). In 209 of 231 deliveries (90.5%), intrapartum ZDV was given (Table 2). The women who did not receive ZDV during labour were diagnosed after delivery, had a very fast delivery, delivered spontaneously at home, or refused treatment. The last mean CD4 cell count before

delivery was reported for 214 births (83.9%) and ranged from 55 to 1268 cells/μL (mean 472 cells/μL; median 444 cells/μL). The last mean CD4 cell count increased from 420 cells/μL in 1994–1999 to 476 cells/μL in 2000–2008 (P=0.06). Of 214 women, 19 (8.9%) had a CD4 cell count of <200 cells/μL (Table 1). Mean CD4 cell count at delivery according to time of ART treatment is shown in Figure 2. It appears that CD4 cell count was significantly higher at delivery when ART was initiated before week 14 of gestation than when it was initiated after week 14 (mean 484 cells/μL vs. 420 cells/μL; P<0.05), which was not explained by the women initiating ART in the third trimester. However, when women on ART at the time of conception were excluded from the analysis, this finding did not remain significant. No linear trend was found. Viral load at delivery was reported for 206 births and in 81% of these the women had undetectable levels of viral HIV RNA.

3) Notably,

3). Notably, Alectinib in vivo qChIP experiments revealed that CtrA occupied the fliF promoter at similar levels in ΔfliG and ΔtipF (99 ± 4% and 80 ± 6% relative to WT, respectively) (Fig. 3), indicating that the increase in class II flagellar gene transcription

in ΔfliG and ΔtipF mutants is not due to an elevated occupancy of CtrA at the promoter(s). Consistent with fliF upregulation seen in ΔfliG and ΔtipF by the β-galactosidase assay, qChIP revealed that the occupancy of FlbD (repressing class II genes) was decreased at the fliF promoter in the ΔfliG (45 ± 1%) and ΔtipF (51 ± 8%) strains (Fig. 4a). FliX, the regulatory factor that links the status of flagellar assembly to FlbD activity (Muir & Gober, 2005), was present at the class II promoters, at higher levels than WT, in ΔfliG (170

± 7%) and ΔtipF (144 ± 4%), consistent with the decreased levels of FlbD at the fliF promoter (Fig. 4a). FliX has been shown to interact with FlbD and block its access to enhancer DNA sequences in vitro (Dutton et al., 2005), and this new qChIP-based approach further suggests that FliX occupies the promoters to modulate FlbD activity at the class II-fliF promoter in vivo. Next, we determined the presence of FlbD and FliX at the class III-flgE and class IV-fljL promoters. qChIP showed that FlbD occupancy at the class III-flgE promoter was reduced in ΔfliG (68 ± 5%) and ΔtipF strains (75 ± 10%) (Fig. 4b), while that of FliX was elevated (155 ± 5% in ΔfliG and 227 ± 9% in ΔtipF) (Fig. 4b). These data demonstrate that the ΔtipF

strain is similar to the ΔfliG mutant strain with regard to the occurrence of FlbD VX-809 molecular weight and FliX at the flgE promoter. It is further consistent with the view that FliX is also present at class III promoters to block FlbD access. The class IV-fljL promoter, however, had an abundance of FlbD similar to WT (123 ± 8%) and decreased levels of FliX (64 ± 7%) in ΔtipF, while the ΔfliG mutant had decreased FlbD (20 ± 2%) and increased FliX (200 ± 9%) (Fig. 4c). These results, also supported by the β-galactosidase promoter-probe assays (Fig. 2), suggest that, unlike FliG, TipF is not necessary to confer the transcription of class IV flagellar genes. Both flbD∷Tn5 and fliX∷Tn5 mutant strains were included as controls. Accordingly, FlbD was considerably Vildagliptin decreased at the fliF (7 ± 1%), flgE (22 ± 3%), and fljL (7 ± 1%) promoters in the flbD∷Tn5 mutant compared with WT (Fig. 4a–c). Similarly, the fliX∷Tn5 mutant had decreased levels of FliX at the fliF (8 ± 2%), flgE (15 ± 1%), and fljL (15 ± 1%) promoters (Fig. 4a–c). The ΔtipN mutant possessed lowered levels of FlbD at the fliF (69 ± 5%) and flgE (57 ± 3%) promoters, while fljL (103 ± 9%) was near WT levels (Fig. 4a–c). FliX was present at the fliF (109 ± 8%), flgE (166 ± 9%), and fljL (129 ± 25%) promoters in the ΔtipN mutant relative to WT. Because the ΔtipN mutant frequently possesses multiple flagella that are often misplaced (Huitema et al., 2006; Lam et al.

Associations between the categories of the supplementary variable

Associations between the categories of the supplementary variable and the others used to build the map are described by interpreting the position of the supplementary categories in relation to the other categories’ position on the map. “
“Increasing numbers of travelers are visiting high altitude locations in the Andes. The epidemiology of acute mountain sickness (AMS) among tourists to high altitude DZNeP in South America is not well understood. A cross-sectional study to evaluate the epidemiology, pre-travel

preparation, and impact of AMS among travelers to Cusco, Peru (3,400 m) was performed at Cusco’s International Airport during June 2010. Foreign travelers, 18 years or older, staying 15 days or less, departing Cusco were invited to participate. Demographic, itinerary, and behavioral

data were collected. The Lake Louise Clinical score (LLCS) was used to assess AMS symptoms. In total, 991 travelers participated, median age 32 years Dasatinib in vivo (interquartile range 25–49), 55.5% female, 86.7% tourists, mostly from the United States (48.2%) and England (8.1%). Most (76.7%) flew from sea level to Cusco and 30.5% visited high altitude in the previous 2 months. Only 29.1% received AMS advice from a physician, 19% recalled advice on acetazolamide. Coca leaf products (62.8%) were used more often than acetazolamide (16.6%) for prevention. AMS was reported by 48.5% and 17.1% had severe AMS. One in five travelers with AMS altered their travel plans. Travelers older than 60 years, with recent high altitude exposure, who visited lower cities in their itinerary, or used acetazolamide were less likely to have AMS. Using coca leaf products was associated with

increased AMS frequency. AMS was common and adversely impacted plans of one in five travelers. Acetazolamide was Resminostat associated with decreased AMS but was prescribed infrequently. Other preventive measures were not associated with a decrease in AMS in this population. Pre-travel preparation was suboptimal. International travel to the South American Andes Mountains has doubled in the past 10 years. Tourist arrivals to Bolivia, Colombia, Ecuador, and Peru went from 2.5 million in 2000 to 6.2 million in 2009.[1] The majority of these tourists visited major cities above the high altitude mark of 2,500 m,[2] like La Paz (3,660 m) in Bolivia, Quito (2,850 m) in Ecuador, and Bogota (2,640 m) in Colombia. Cusco (3,400 m), in the south Andes of Peru, is a major tourist destination visited by over 1 million foreign tourists in 2008.[3] Of note, most tourists ascend to Cusco in flights departing at sea level and lasting less than 1 hour. Short-term exposure to high altitude is associated with acute mountain sickness (AMS), a common and usually self-limited illness.[2] In a prior survey of travelers to Cusco, AMS was as common as traveler’s diarrhea.

We recorded the responses of superficial dorsal horn neurons in m

We recorded the responses of superficial dorsal horn neurons in mice to intradermal injection of the pruritogens chloroquine and histamine. Scratching within an area 5–17 mm distant from the injection site, outside of the units’ mechanoreceptive fields (off-site), Epacadostat cell line significantly inhibited chloroquine-evoked and histamine-evoked responses without affecting capsaicin-evoked firing. This is consistent with observations that scratching at a distance from a site of itch is antipruritic. In contrast, scratching directly at the injection site (within the receptive field; on-site) had no effect on chloroquine-evoked neuronal firing, but enhanced the same neurons’

responses to intradermal injection of the algogen capsaicin. Moreover, neuronal responses to histamine were enhanced during on-site scratching, and this was followed by suppression of firing below baseline levels after termination of scratching. Scratching thus inhibits pruritogen-responsive neurons in a manner that

depends on the input modality (i.e. pain vs. histamine-dependent or histamine-independent itch) and selleck chemicals llc skin location. “
“Involvement of fronto-parietal structures within the right hemisphere in bodily self recognition has gained convergent support from behavioural, neuropsychological and neuroimaging studies. Increases in corticospinal excitability via transcranial magnetic stimulation (TMS) also testify to right hemisphere self-related processing. However, evidence for self-dependent modulations of motor excitability is limited to the processing of face-related information that,

by definition, conveys someone’s identity. Here we tested the hypothesis that vision of one’s own hand, as compared with vision of somebody else’s hand, would also engage specific self-hand processing in the right hemisphere. Healthy participants were submitted to a classic TMS paradigm to assess changes in corticospinal excitability of the right (Experiment 1) and left (Experiment 2) motor cortex, while viewing pictures of a (contralateral) still hand, which could either be their own (Self) or not (Other). As a control for body selectivity, subjects were also presented with pictures of a hand-related, but non-corporeal object, i.e. a mobile phone, which could similarly be their own or not. Results showed a selective Farnesyltransferase right hemisphere increase in corticospinal excitability with self-hand and self-phone stimuli with respect to Other stimuli. Such a Self vs. Other modulation of primary motor cortex appeared at 600 ms and was maintained at 900 ms, but was not present at earlier timings (100 and 300 ms) and was completely absent following stimulation of the left hemisphere. A similar pattern observed for self-hand and self-phone stimuli suggests that owned hands and objects may undergo similar self-processing, possibly via a different cortical network from that responsible for self-face processing.

3% (treatment initiated 14–28 weeks, non-breastfeeding, low CS ra

3% (treatment initiated 14–28 weeks, non-breastfeeding, low CS rate, baseline Talazoparib datasheet CD4 cell count >200 cells/μL) [61], a 50% reduction in a Thai study to 9.4% (mean treatment only 25 days and oral zidovudine during labour) [130]; 0.8% transmission

for women treated with zidovudine monotherapy and assigned to PLCS in the Mode of Delivery study [131]. Since 2000, BHIVA guidelines have recommended zidovudine monotherapy plus PLCS for women with CD4 cell counts above the prescribed threshold for initiating HAART and with an untreated VL <10 000 HIV RNA copies/mL plasma, based on these and other data and on the published relationship between VL and transmission [132]. No transmissions were observed in the UK and Ireland among the 464 pregnancies managed by zidovudine monotherapy and PLCS between 2000 and 2006 reported to the NSHPC.

The median delivery VL in these women was 400 (IQR 61–1992) HIV RNA copies/mL [4]. 5.3.5 Women who do not require treatment for themselves should commence temporary HAART at the start of the second trimester if the baseline VL is >30 000 HIV RNA copies/mL plasma. (Consider starting earlier if VL > 100 000 HIV RNA copies/mL.) Grading: 1C VL data also influence recommendations relating to mode of delivery Selleck GSK 3 inhibitor (see below). Major determinants of the probability of achieving a VL <50 HIV RNA copies/mL plasma by the time of delivery are the baseline untreated VL and the time available to achieve this target. In the Mma Bana study, VLs <400 HIV RNA copies/mL plasma were achieved by the time of delivery in 96% (lopinavir/ritonavir-based) to 100% (abacavir/lamivudine/zidovudine) of mothers with baseline VL <1000 HIV RNA copies/mL plasma and in 86% (lopinavir/ritonavir-based) to 90% (abacavir/lamivudine/zidovudine) if baseline VL >100 000 HIV RNA copies/mL. When therapy was initiated at 31–34 weeks, only 78% of mothers on PI-based therapy had achieved this target [66]. Data from a UK multicentre study retrospectively analysing therapy outcomes in pregnant women initiating HAART at a median

gestation of 23 weeks demonstrate very low rates of complete suppression in women with a baseline VL in the upper quartile (>32 641 HIV RNA copies/mL) Alanine-glyoxylate transaminase with only 46% achieving <50 HIV RNA copies/mL by 36 weeks' gestation (the data point used to make most delivery management decisions) and this fell to 37% for VLs >100 000 HIV RNA copies/mL [133]. For all VLs >10 000 HIV RNA copies/mL, treatment initiation later than 20.3 weeks’ gestation was associated with significantly less likelihood of successful VL suppression. To address this, the Writing Group recommend that HAART should be commenced at the start of the second trimester, or as soon as possible thereafter, in women with a baseline VL >30 000 HIV RNA copies/mL plasma. 5.4.1 A woman who presents after 28 weeks should commence HAART without delay.


subjects were seated in a chair in a magnetically shi


subjects were seated in a chair in a magnetically shielded room and listened to the group sequence and the random sequence in separate sessions by using a magnetoencephalography (MEG)-compatible earphone connected with silicon air tubes. Before the experiment, we asked the subjects whether the tones could be heard from both ears with the same loudness and all of them reported that they could be heard correctly. The group sequence was always presented in a session after the random sequence in order to avoid the interference of grouping effect on the random sequence. While the subjects were listening, they were instructed to press a button by their right index finger if they noticed the omission of a tone. Because the total length of the group sequence was long (over 20 min), we divided this into two sessions. Thus, the experiment consisted of one session of random see more sequence (8 min) and two sessions of group sequence (12 min × 2). After each session, in order to check the subjects’ arousal level and fatigue, we asked the subjects whether they felt sleepy or wanted to have

a short break. If the subjects felt tired, Oligomycin A cost they were allowed to have a short break. Including the short breaks, the total time of the measurement was about 35–40 min. The experimental design was a two-way mixed design with musical experience (musicians or non-musicians) and omission (random, within-group or between-group). At the end of the experiment, we asked the subjects whether the stimuli in the group sequence had been recognised as the LLS pattern and all subjects reported noticing this pattern. The auditory evoked magnetic fields were recorded with a 306-channel whole-head MEG system (Vectorview, Elekta Neuromag Oy, Finland), which contained 102 sensor triplets consisting of two planar gradiometers and one magnetometer each. The exact head position with respect to the sensors was determined by measuring signals from four indicator coils attached

to the scalp. In addition, three head landmarks (the nasion and bilateral pre-auricular points) and the subject’s head shape were recorded with a spatial digitiser (Polhemus Inc., Colchester, USA) before Cyclin-dependent kinase 3 the experiment. These data were used for co-registration with the individual structural magnetic resonance imaging data obtained using a 0.2 T magnetic resonance scanner (Signa profile, GE Health Care, Waukesha, USA). The MEG data were recorded with a bandpass filter of 0.1–200.0 Hz and a sampling rate of 600.0 Hz. To reduce external noise, we used spatio-temporal signal space separation methods (MaxFilter, Elekta Neuromag Oy) with a correlation window of 900 s, which covered the whole length of each session, and a correlation limit of 0.980. The acquired data were low-pass filtered using a fifth-order Butterworth zero-phase filter with a cut-off frequency of 40 Hz.

There are concerns about the development of three children: two w

There are concerns about the development of three children: two with speech delay and one who is failing to thrive. Two children are known to have been fostered. All 30 young women included in this study had been independently Alisertib in vivo notified through routine systems

to the NSHPC. Twenty-seven were reported as paediatric cases (eight born in the British Isles and 19 born abroad) and three (all born abroad) when pregnant at 16 years or older. All 21 live births had also been notified to the NSHPC, but 15 of the 21 miscarriages and terminations had not. In the UK and Ireland, young women infected with HIV perinatally or in early childhood are now becoming sexually active and having children of their own. This cohort shares common characteristics with small cohorts of perinatally infected pregnant young women reported from Europe [5], the USA [6, 7], Puerto Rico [6] and India [7]; these include significant rates of unplanned pregnancy, low rates of MTCT despite archived resistance mutations limiting treatment options, inconsistent adherence to

cART complicating management in pregnancy, and complex social circumstances. Among the young women aged 12 years and over receiving care in STA-9090 clinical trial 21 participating clinics, 12% were known to have had at least one pregnancy, with a 14% first-trimester miscarriage rate, lower than the 24% reported in horizontally infected women [8], although this could be an underestimate as a result of likely under-reporting of early pregnancy loss. In the USA, which has the largest published cohort of 638 perinatally infected young women, the cumulative incidence Tacrolimus (FK506) of first pregnancy by 19 years of age was 17.2% [95% confidence interval (CI) 11.1, 23.2], substantially lower than first-time pregnancy rates in US girls of a similar age who were presumed to be HIV uninfected

(33.5 per 1000 person-years vs. 86.7 per 1000 person-years, respectively). The authors speculated that this might be attributable to increased contraceptive availability and awareness, or reduced fertility, in HIV-infected adolescents compared with the general population. They reported that sexually active girls had a higher VL and a lower CD4 percentage and were less likely to be on cART than those who were not sexually active [9]. In a recently reported cohort study of 67 pregnancies in 58 predominantly horizontally infected UK teenagers (median age at conception 18 years), 82% of pregnancies were unplanned, 58% delivered with undetectable virus and one infant was infected. Two-thirds of this cohort were newly diagnosed with HIV during antenatal screening, and therefore had not had prior access to HIV-related sexual and reproductive health support. Despite subsequent access to clinical care and contraceptive services, almost a quarter were pregnant again within 1 year and post termination/delivery contraceptive use was suboptimal [10].

The pharmacist also ensures that information on medication change

The pharmacist also ensures that information on medication changed, started and stopped is documented. PTTAs are

currently not screened by a second pharmacist but should be checked by the doctor. Anecdotal evidence is that this does not happen routinely. 80% of all weekday discharge medication lists are PTTAs. This study aimed to assess a representative sample of PTTAs for safety (error rate) and quality of documentation. This was a retrospective study. Data collection took place on single days during seven convenient, non-consecutive weeks between October 2013 and January 2014. Stratified sampling (proportionate allocation) was used to ensure appropriate representation of all clinical specialties. The data collection tool was based on a previous similar study (Linda Dodds, see more personal communication, Smad inhibitor 2013), piloted by pre-registration pharmacists and pilot data validated by a senior clinical pharmacist. Pre-registration pharmacists collected final versions of PTTAs written a week before the data collection day and documented the specialty, the medicines from the drug history, inpatient chart and the PTTA. They noted any differences between the three lists and the documentation of such. Senior clinical pharmacists assessed the

discrepancies between the lists to determine intentional and unintentional changes, and the quality of documentation. Ethics approval was not needed as this was a service evaluation. Data was entered into MS Excel for analysis. Four hundred twenty-eight PTTAs were reviewed. All could be assessed for errors. Errors were found for 12/428 patients. (2.8%, 95% CI 1.3%–4.3%). Sixty-nine PTTAs were not evaluated for documentation of changes. Fifty-four PTTAs from the Women’s and Children’s wards did not have this information available at the time of data collection. Fifteen

patients had no changes to their medication. 272/359 (75.8%, 95% CI 71.5–81.3%) patients were discharged with all relevant information regarding medication changes documented in the DN. The most serious error was in a surgical patient who was taking a high dose of oral morphine sulphate plus tramadol daily before discharge but was discharged without a strong opiate. Other errors included an incident of therapeutic duplication (antibiotics) and analgesics and anti-emetics PKC inhibitor missing from PTTAs despite being taken regularly just before discharge. Two point four per cent error rate on pharmacist-written discharge medication lists is remarkably low compared to the literature for traditional DNs. Additionally, 76% of DNs had complete information regarding medications initiated and stopped. Dodds showed that two-thirds of doctor-written discharge summaries were inaccurate prior to a pharmacy check.1 Our PTTAs can be improved further as not providing information on medication changes to primary care and community colleagues can give rise to errors and adverse events after discharge.

Furthermore, there is evidence that transcripts can be cleaved wi

Furthermore, there is evidence that transcripts can be cleaved within the A-site of paused or stalled ribosome (Hayes & Sauer, 2003) and such cleavage may lead to the triggering of trans-translation (Moore & Sauer, 2007). Thus, the role of trans-translation in reducing the effects of antimicrobial agents may relate more to overcoming the consequences of translational errors and truncated mRNA than to the stalled state caused by direct ribosome inhibition. As well as exposure to antimicrobial agents, there are several reports indicating that tmRNA levels can increase

under other conditions. For instance, increased levels of tmRNA correlated with G1–S transition in the cell cycle in Caulobacter crescentus

(Keiler & Shapiro, 2003; Hong et al., 2005) and in response to heat or chemical stress in Bacillus subtilis (Muto et al., 2000). In the former study (Keiler & Shapiro, 2003), the changing level of tmRNA GSK2126458 supplier was believed to be critical to the timing of the selleck cell cycle. In bacteria, mature tmRNA is one of the most abundant RNA species. tmRNA levels in M. smegmatis are equivalent to those reported for E. coli (Lee et al., 1978; Moore & Sauer, 2005). The abundance of tmRNA is a likely consequence of a high rate of trans-translation; for instance, approximately 0.4% of translation reactions in E. coli are terminated by trans-translation (Moore & Sauer, 2005). The abundance of tmRNA is also likely a consequence of its stability, which is believed to result from its binding to SmpB (Keiler et al., 2000; Moore & Sauer, 2005) and it is assumed that the majority of mature tmRNA and SmpB is in complex (Keiler, 2008). The half-life of mycobacterial tmRNA under conditions inhibiting RNA synthesis was similar to that reported for Caulobacter sp. swarmer cells and E. coli (Keiler & Shapiro, 2003). The stability of mycobacterial tmRNA was somewhat paradoxical in light of the high level of ssrA promoter Obatoclax Mesylate (GX15-070) activity indicated by the results presented here. However, a previous study of the ssrA promoter of

C. crescentus also indicated that it was one of the most active promoters even under conditions where tmRNA was highly stable (Keiler & Shapiro, 2003). Irrespective of whether high-level ssrA promoter activity maintains tmRNA levels in the absence of ribosome inhibition, the evidence indicated that drug-associated increased levels of tmRNA were the result of increased promoter activity. This interpretation was supported not only by the promoter analysis but also by the finding that tmRNA loss was not affected by the drug exposure. The results presented here indicate that ribosome inhibitors, such as erythromycin, increase the synthesis of tmRNA in mycobacteria and thus provide an underlying mechanism for the increased levels of tmRNA following exposure to such agents.

56; SmartGene, Zug, Switzerland) which contains all genotypic HI

5.6; SmartGene, Zug, Switzerland) which contains all genotypic HIV resistance tests performed by the four authorized laboratories in Switzerland [19]. A GSS was defined for each NRTI, NNRTI and PI using the Stanford algorithm (version 6.0.3), such that 0 denotes full resistance to a given drug, 0.5 denotes intermediate resistance C646 supplier and 1 denotes full susceptibility. Raltegravir and enfuvirtide were deemed fully susceptible if no mutation

in the International AIDS Society (IAS)-USA mutation list was detected in integrase and glycoprotein 41 (gp41) tests, respectively [20]; or in the absence of these tests, full susceptibility was assumed for these drugs (and for maraviroc) unless these drugs had already been used in a failed regimen. To derive an overall GSS for therapy, we summed the scores of each drug in the regimen. We also considered a number of alternatives to selleckchem this overall GSS, to see if these alternatives suggest some simple rules for clinical practice. First, we replaced the overall GSS with two components – a GSS for darunavir and a GSS for background therapy. Secondly, we considered whether each of these component GSS values can be approximated by simple clinical

measures. As rough measures of existing resistance to darunavir, we assessed whether the patient failed on both amprenavir and saquinavir and counted the number of failed PI regimens. As rough measures TCL of the potency of background therapy, we assessed whether the patient had at least one other second generation antiretroviral in the regimen in addition to darunavir and counted the number of de novo drugs in the regimen in addition to darunavir. With limited data for analysis, we took a Bayesian approach to fitting Cox proportional hazards models for time to virological failure. Given

that we assessed failure in each of three periods, we used a discrete time version of the Cox model with an offset that adjusts for variation in the time between assessments [21]. For each predictor in our model, we asserted a ‘vaguely informative’ prior where ‘the percentiles of the prior distribution would be viewed as at least reasonable if not liberally inclusive by all those working in the research topic’ [22]. Each prior was represented by a lognormal distribution for a hazard ratio, data that reproduced this distribution were added to the observed data, and standard software was then used to estimate an approximate posterior hazard ratio by a weighted averaging over observed and prior data with each set of prior data assigned to a separate stratum [23]. A priori, we classified each predictor into one of five categories. First we rescaled continuous predictors age, viral load and CD4 cell count into clinically meaningful units (per 10 years, log10 copies and 100 cells/μL, respectively) and centred each about its median.