There were 37 studies (n=992) on UPPP, of which 29 (n=819) contained sufficient patient data to allow the response rate to be calculated, and 17 contained individual patient data. To this last group was added 2 publications, which were duplicates of studies in the other 2 groups, but which now contained individual patient data (n=345).
There was 1 study (n=20) of nasal operations.
There was 1 study (n=20) of UPPGP.
There were 2 studies (n=34) of LMG.
There was 1 summary of 9 studies (by the same authors) of GAHM (n=55), plus one further report (n=9) which was not included in the meta-analysis.
There were 4 studies (n=489) of MMO and advancement.
There were 3 studies (n=99) of tracheotomy.
Nasal operations: there were no significant differences between the baseline and post-operative values of RDI in the one study reviewed.
UPPP operations: changes in AI and RDI were highly significant (p<0.0001). The weighted Pearson correlation coefficient (20 of the 37 papers) between the mean baseline AI and percentage change in AI was 0.502 (p=0.024), suggesting that a high baseline AI was associated with a lower percentage decrease in AI. The correlation between the baseline RDI and the percentage change in RDI (18 papers) was 0.106.
The response rate was 65.8% (129 out of 196 patients) when defined as a 50% drop in AI.
The response rate was 52.8% (114 out of 216 patients) when defined as a 50% drop in RDI.
The response rate was 40.7% (137 out of 337 patients) when defined as a 50% drop in AI or RDI, with consequent achievement of an AI of less than 10 or an RDI of less than 20.
The Spearman's rank correlation coefficients showed that patients with high baseline AIs tended to have smaller percentage reductions in AI (n=196, correlation -0.202, p=0.005) and greater reductions in RDI (n=80, correlation 0.249, p=0.026).
Information on the location of pharyngeal narrowing or collapse, and complications of UPPP, was also reported.
UPPGP operations: there was a mean percentage decrease in RDI of 42.6% (no p-value given). Ten of the 20 patients had at least a 50% reduction in RDI.
A further 19 patients had a response rate of 67% at 6 months, when response was defined as a 50% reduction in AI.
LMG operations: the response rate was 41.7% (5 out of 12 patients) when defined as a 50% reduction in RDI.
Lingualplasty operations: the response rate was 77% (17 out of 22 patients) when defined as a post-operative RDI of less than 20 per minute and at least a 50% reduction from pre-operative RDI.
GAHM operations: only 6 out of 55 patients had GAHM alone; 49 out of 55 patients in this group had UPPP and GAHM, either synchronously or at different times.
The response rate was 67% (37 out of 55 patients) when defined as a combination of a post-operative RDI of 20 or less, at least a 50% reduction from pre-operative RDI and minimal oxygen desaturation.
MMO operations: the mean pre- and post-operative RDIs were 72.0 (plus or minus 25.8) and 8.8 (plus or minus 6.1), respectively (n=30).
In another study, the response rate was 61.1% (146 out of 239 patients) when defined as a RDI of less than 20 at 6 months, at least a 50% reduction from pre-operative RDI, and lowest oxygen saturation equivalent to that of a second night of continuous positive airway pressure titration. The likelihood of response tended to diminish with increasing pre-operative apn0ea severity.
In a third study, the response rate (defined as a RDI of less than 10 at 6 months) was 20% (1 out of 5 patients), rising to 77.8% (14 out of 18 patients) when at least one adjunctive procedure was used and 100% (7 out of 7 patients) when UPPP was used as an adjunct.
In a fourth study, the response rate was 95.2% (20 out of 21 patients) when defined as a post-operative RDI of less than 10.
Tracheotomy: in a series of 11 patients, excessive daytime sleepiness and sleep disruption were resolved in 9 of the patients; the mean follow-up was 17 months. The data on the AI or RDI were not given.
In a series of 50 patients (mean follow-up 32 months), all patients achieved an AI of less than 5; pre-operatively, all AIs were greater than 50.
In a third series of 38 patients, excessive daytime sleepiness was resolved in all patients within 48 hours post-operatively. No data were given on the AI or RDI values.