Table 4.
Sensitivity analyses for effect of sample non-response on estimate of dementia prevalence in 2011
|
Total |
Men |
Women |
Cambridgeshire | Newcastle | Nottingham | ||||
|---|---|---|---|---|---|---|---|---|---|
| Prevalence | Number | Prevalence | Number | Prevalence | Number | ||||
| Main analysis | 6·5% | 670 000 | 4·9% | 227 000 | 7·7% | 443 000 | 6·1% | 7·2% | 6·0% |
| Dynamic population | 6·6% | 685 000 | 5·0% | 230 000 | 7·9% | 455 000 | 6·3% | 7·3% | 6·1% |
| All in care not seen have dementia | 5·9% | 598 000 | 4·8% | 222 000 | 6·8% | 396 000 | 5·4% | 6·8% | 5·6% |
| Simple MAR* | 6·0% | 625 000 | 4·8% | 221 500 | 7·0% | 404 000 | 5·3% | 7·1% | 5·5% |
| Type of refusal MAR* | 6·0% | 625 000 | 4·8% | 221 000 | 7·0% | 403 500 | 5·3% | 7·1% | 5·5% |
| All refusals at 50% increased risk of dementia* | 7·7% | 797 000 | 6·1% | 279 000 | 8·8% | 516 500 | 6·7% | 9·0% | 7·2% |
| All refusals at 100% increased risk of dementia* | 8·5% | 882 000 | 6·7% | 309 000 | 9·9% | 572 000 | 7·4% | 10·0% | 8·0% |
| Ill health at 100% increased risk, passive or proxy refusals at 50% increased risk, active refusals at 20% decreased risk* | 6·9% | 717 000 | 5·5% | 253 000 | 8·0% | 464 000 | 6·1% | 8·1% | 6·4% |
Descriptions of scenarios are provided in Methods and appendix. MAR=missing at random.
*
Bayesian model.