what can an elderly person with life-long insomnia try, to finally get a few years of decent sleep

The short version: for lifelong insomnia in old age, the best-evidenced thing is behavioral therapy, and the honest gain is modest, not a cure.

Cognitive behavioral therapy for insomnia (CBT-I) is the one approach with randomized trials in people over 60. A systematic review of those trials found a mild benefit: sleep-maintenance therapy cut time awake after falling asleep by roughly 22 to 24 minutes, and one trial added about 14.6 minutes of total sleep by diary [1]. That is real but small. Bright light and exercise had limited evidence in the same review, and no bright-light trial met its inclusion criteria [1].

The drug evidence is thinner than it looks. Low-dose doxepin (6 mg) beat placebo over four weeks in older adults with chronic insomnia, improving sleep maintenance and total sleep time [2]. Four weeks is not lifelong use, and the gains were modest. A 2025 network meta-analysis of 102 trials found most insomnia drugs raised the risk of drowsiness, dizziness, or headache versus placebo, and doxepin was the exception, with lower headache risk [3]. Suvorexant has a trial in older adults with Alzheimer's and insomnia showing better sleep maintenance than placebo [4][5], but that is a different population than a cognitively intact elder.

What this evidence cannot see: no trial here followed lifelong insomniacs for years, and none compared CBT-I head to head against drugs over the long haul. The behavioral approach is the safer bet because the harms are minimal, not because the benefit is large.

My call: start with CBT-I, add low-dose doxepin only if sleep maintenance stays broken, and expect incremental gains, not transformation. Confidence: moderate on CBT-I being first-line, low on any single drug delivering years of good sleep.