Blood-Pressure Medication in Older Adults: What OPTiMISE Found

What happens when an older person with controlled blood pressure stops one of several blood-pressure medicines? The OPTiMISE trial tested a monitored reduction in a selected group of people aged 80 or older.

Researchers first checked blood-pressure control after 12 weeks, then used health records to follow hospital admissions and deaths for a median of four years. The two reports answer different questions: whether control can be maintained in the short term, and what happened to clinical outcomes over a longer period.

The original trial enrolled 569 people aged 80 or older from 69 general practices. They had systolic blood pressure below 150 mm Hg and were taking at least two blood-pressure medicines. Their own primary care physicians considered them suitable for a reduction. This was a selected group, rather than a sample of everyone with hypertension.

Participants were randomly assigned either to have one blood-pressure medicine removed or to continue usual care. The reduction group had a safety appointment at four weeks, and the protocol allowed treatment to be put back. Both groups had a 12-week assessment.

That sequence matters. The intervention included selection, withdrawal and a check on what happened next. Taking the tablet away was only one part of the process.

At 12 weeks, 86.4% of people assessed in the reduction group and 87.7% in usual care had systolic blood pressure below 150 mm Hg. The trial met its planned statistical test for preserving that level of control. Average systolic pressure nevertheless rose more in the reduction group. The 2020 report therefore answered a short-term question about blood-pressure control; it could not settle the longer-term balance of benefits and harms.

The 2024 follow-up used health records to examine hospital admission or death over a median of four years, with main outcome data for 564 participants.

Hospital admission or death occurred in 202 of 280 participants in the reduction group and 218 of 284 in usual care: about 72% and 77%. The adjusted hazard ratio was 0.93 (95% confidence interval, 0.76–1.12). The study did not detect a statistically significant difference in this combined outcome.

The range around that estimate is important. It includes the possibility of lower risk and the possibility of some higher risk. The two percentages alone cannot establish that reducing treatment prevented hospital admissions or deaths. They also combine two events, so they are not mortality rates.

The study also reported a per-protocol estimate: a hazard ratio of 0.80 (95% confidence interval, 0.64–1.00). That analysis excluded reduction-group participants who had not reduced treatment or had restarted medication in the first 12 weeks. The follow-up paper reports a p value of 0.053.

Keeping people in their originally assigned groups preserves the initial randomized comparison. Selecting people according to what happened afterward can introduce bias, as a separate methods analysis illustrates. The lower estimate is worth seeing, but it cannot simply replace the main result as proof of a survival benefit.

The researchers describe limits: the follow-up was not designed to establish non-inferiority for long-term clinical outcomes. Rare events need larger studies. Routine records also could not answer every question about how participants felt after treatment changed.

The findings on hospital admission and death cannot, by themselves, establish improvements in symptoms or daily function.

OPTiMISE tested the removal of one medicine, with a plan to check the result and restore treatment when needed. That full sequence remains central to its findings.