Pain relief creams containing drugs similar to aspirin stop working in patients with osteoarthritis (OA) within weeks, research shows.
OA affects joints like the hips and knees
Current OA guidelines recommending lotions containing non-steroidal anti-inflammatory drugs should be revised, say the authors.
The Nottingham University scientists found the treatment was no better than a dummy cream after two weeks.
Their findings are reported in the British Medical Journal.
OA is the most common form of arthritis and a major cause of disability in elderly people. About 80% of 80 year olds have it.
NSAIDs, such as ibuprofen, are commonly used to treat the pain associated with this condition, but few trials have looked at their effects for longer than two weeks, according to Dr Weiga Zhang and colleagues.
They analysed 13 trials comparing topical NSAIDs (treatments applied to the skin) with dummy treatment or NSAIDs taken in a tablet form in nearly 2,000 patients with OA.
Topical NSAIDs provided better pain relief than the dummy treatment in the first two weeks, but after a fortnight they were no better than the fake treatment.
They were less effective than comparable oral versions in the first two weeks of treatment and caused more skin-related side effects such as itch, rash and burning.
Their relief of stiffness also waned after two weeks.
In comparison, oral NSAIDs were an effective long term treatment.
The researchers said guidelines supporting the use of topical NSAIDs for OA should be revised.
Lead author Weiya Zhang said: "I would hope in the future when we update the guidelines we would say current evidence only provides support for the use of the drug for the short term period.
"I do not think it is a suitable treatment for OA because OA is a chronic condition and requires long term treatment.
"For acute pain, like pain due to a sport injury, topical NSAIDs might be useful," he said.
He called for further long term studies over months rather than weeks.
Dr Peter Dawes, consultant rheumatologist and honorary secretary of the British Society of Rheumatology, said: "To show good benefit you have got to show there's a long term benefit.
"Guidelines are based on the short term data.
"If patients are not getting a long term benefit then they need to have a proper assessment and a look at what are the other forms of treatment that can be used."
He said OA might not always be the thing causing the pain.
"If you are having symptoms from arthritis and it's not settling down you may need to have a specialist opinion," he said.
A spokeswoman from Arthritis Research Campaign said: "We certainly welcome this study which has shown that topical NSAIDs are only of use in helping OA pain over a relatively short time period and are not effective in the longer term.
"It is useful to question the widespread use of NSAIDs, both topical and oral, when, often, a more simple regime of paracetamol or paracetamol and codeine provides perfectly adequate and safer pain relief for many patients with OA," she said.