Daniel Grint

International Statistics and Epidemiology Group | LSHTM Collaborating with international research partners to improve health in low and middle-income countries

📢Funded fellowships available📢

We have two fully funded medical statistics fellowships available for applicants from sub-Saharan Africa, covering

▶️1 yr MSc in Medical Statistics at LSHTM
▶️1 yr placement at a partnering research institution
🗓️Closing date 31st March

www.lshtm.ac.uk/research/cen...

I'm currently jumping through hoops at one journal where the editors seem incapable of 'editing' and every single revision, no matter how minor, goes back out to the reviewers. 8 months in the system and counting...

It's an excellent paper 😄

I write my abstract in haiku.

Do no harm - re-evaluating the risks of overtreatment in community-wide tuberculosis screening Background Community-wide screening is a crucial strategy to end tuberculosis (TB), but a common concern is potential harm from overtreatment following false positive diagnoses. However, current refer...

In a new preprint (NOT PEER REVIEWED) we continue to explore the challenge of overtreatment in community screening. It seems the benefits of TB treatment far outweigh the harms, especially once we accept that sputum culture is not perfect. Important food for thought. www.medrxiv.org/content/10.1...

For this reason the combination of Xpert CT and a detailed CXR examination seems to be best at identifying those who may relapse post treatment.

They may all represent seperate pathways, except smear and CT which intuitively may be related based on sputum. The great benefit of Xpert CT is the automatic readout with no need for reader interpretation. I imagine much of the variability in culture/smear comes from the operator.

Correlations between CT, culture and CXR were weak, it's only smear that was reasonably correlated with CT. Smear and culture were weakly correlated with each other (R^2 14%), but neither correlated with CXR (both R^2 <10%).

Hi Gabriele
Ultra is worse than Xpert at differentiating between the higher levels of disease severity. The study population was skewed towards a higher degree of disease severity, which may explain the lack of correlation between Ultra and CXR.

Xpert MTB/RIF® cycle threshold as a marker of TB disease severity; Implications for TB treatment stratification

✅ Just Accepted
#IDSky
🔗 https://bit.ly/3WbLI1G

Push to main.

This paper has now been peer reviewed and is available open access at CID.

Acetaminophen Use During Pregnancy and Children’s Risk of Autism, ADHD, and Intellectual Disability This nationwide cohort study with sibling control analysis examines the association of acetaminophen use during pregnancy with children’s risk of autism, ADHD, and intellectual disability.

apropos of nothing

Body mass index and tuberculosis risk: an updated systematic literature review and dose–response meta-analysis AbstractBackground. The relationship between nutritional status and tuberculosis is critically important but poorly understood. We extended a 2009 review c

Nutritional status is a key determinant of TB. In this updated systematic review and dose-response meta-analysis (43 cohorts inc > 26 million people) we redefined the relationship between body mass index (BMI) and TB. doi.org/10.1093/ije/...

The potential impact, cost and cost-effectiveness of tuberculosis interventions - a modelling exercise Background While a range of interventions exist for tuberculosis prevention, screening, diagnosis, and treatment, their potential population impact and cost-effectiveness are seldom directly compared,...

At a time when global resources for #TB research, development, prevention, and care are limited, which interventions should be prioritised in high burden settings?

📢 Our new pre-print compares the impact, cost, and cost-effectiveness of 9 TB interventions in 3 countries. 1/n

bit.ly/3V3vXth

"Nobody suspects the dishwasher" sticks in memory from a previous round I attended somewhere.

Academia may not give you job security, flexibility, or wealth, but it will let you unexpectedly connect to eduroam in foreign cities

Article on BBC news. Title: AI designs antibiotics for gonorrhoea and MRSA superbugs Description: Two new potential drugs have been designed by AI to kill drug-resistant bacteria, in a major Massachusetts Institute of Technology study.

I really dislike how science has started calling almost any fancy computational technique AI. 🧪

The framing of this entire article makes it sound like a benevolent AI independently made these drugs.

That is *pure fantasy*.

Instead: a team of scientists made a machine learning model for a study.

p.s. I've been consistently spelling injectAble wrong for 3+ years now.

The front page of the preprint article, showing title and authors

The abstract of the preprint

Forest plot showing primary and secondary trial outcomes, with non-inferiority margins

⚡️New preprint (not peer-reviewed)⚡️

In the SaDAPT trial with people with #HIV and #tuberculosis symptoms in 🇲🇼🇱🇸, we investigated if antiretroviral therapy should be initiated immediately, or delayed until TB results available. ssrn.com/abstract=523...

Outcomes (incl. IRIS) similar across both arms.

Delighted to see our long-acting injectible ART trial (IMPALA) presented at #IAS2025. 2-monthly injections are non-inferior to daily pill taking and overwhelmingly preferred. Next challenge is expanding access to LA ART; excellent news the WHO has recommended LA CAB+RPV in treatment guidelines.

Yeah, that part could definitely be better written. 10% mortality reduction is impressive!

Interpretation beyond that is for the reader. It's possible these results could still change guidelines.

I think you'd probably be right to, assuming there are no potential ill effects. 'Statistical significant' i.e. <0.05 is less of a thing now, but you still have to respect the study design. In this case, they just missed the pre-defined superiority criteria and must report on those lines.

You don't need a Bayesian analysis to be less rigid in interpreting p- values. You're right the result looks promising, but it requires further study.

Mandating culture/PCR for everyone in a vaccine trial makes sense to me, I didn't realise that wasn't the case. But it could (likely will) end up costing more even if the sample size is lower.

Very interesting. In TB treatment trials, it's standard practice to test culture from everyone every few months regardless of symptoms. Consecutive positives define TB relapse without worrying about symptoms. However, this is a big cost driver!!!

Describing the hawkeye automated line calling at Wimbledon as 'AI' can seriously get in the sea.

add 10% to the quote you see when shopping around is my philosophy once it's been through the booking agent

Until there is a test for TB infection with higher specificity than IGRA, the use of the non-inferiority trial design for TB vaccine prevention of infection outcomes is not recommended.

@richardwhite321.bsky.social @cfmcquaid.bsky.social @reinhouben.bsky.social