You are absolutely correct in your apprehension to believe the results.
The smaller pixels (photosites) do not provide better performance than larger pixels, and that is where the results have to be understood. The score DxO gave for the D800/D800e is not based on what the sensor does, it is based on what you can achieve when downsampling the image to ~25% of its original size. DxO evaluates performance on an 8mp image, therefore just about any camera above 8mp will show better results than an 8mp camera, and certainly better than its predecessor. In the scoring system you also have to understand that DxO awards the score based on the highest achievable performance score - even if the camera only performs to that level at one ISO setting and performs worse at every other setting - see a comparison of D3x and D3s where the D3s excels above the D3x in every criteria except at ISO 100, which is where the D3x achieved its score. If you were to fairly compare the maximum quality output of cameras against any other you would either do that at the sensor level or at the same downsample ratio. How about a comparison of the D800 downsampled to 8mp, against the D3s downsampled to 3mp - that would be a fair comparison of output given that there are certainly just as many quarter page magazine prints (3mp) as there are 8x12 paper prints (8mp) being used in the industry..
In order to see the real performance you have to select the little toggle in the upper left of the measurement page labeled [Screen] where you will then see what the performance is at the sensor level. You may be asking why this is important, I mean some would be saying 'the camera I want just got the highest score ever, that is good enough for me' but as is to be expected a considerable percentage of buyers (those without real photography experience or knowledge of composition, lens selection, exposure, aperture, etc) will be looking to crop into all those megapixels hoping to get the image they should have captured correctly to begin with - and that is where the performance is the worst, at the pixel level, or the closer you get to the pixel level. If you compare the performance of the D800 against any Nikon above, say the D80 or so, you will see that its performance at full resolution is no better than most of Nikon's lower megapixel models - and actually begins trailing behind most of them once you get past ~200 ISO.
The relevance of the score has more to do with your intended/needed output. For studio work the D800 should be great as most of those images are headed toward magazine and smaller prints (where the downsample will provide benefit, either digitally or in print) however the other group Nikon has marketed the D800 towards, Landscape photography, will not see the performance they have paid for as their output is typically larger prints needing higher megapixels to yield the extra detail - the area where the D800 performs no better than most of Nikon's bodies other than megapixels. As for walk-around/general photography, like you do with a D700, the D800 is not for that group. Not that it cannot take general subject photographs, but how many of you in that group are looking to incorporate professional level post processing, lenses, and additional overhead all that comes with just to yield the results you are already getting from whatever you already use? Remember too not to be mislead by marketing, you do not need high megapixels to upscale an image.. Upscaling an image allows more pixels to draw the same detail present, which means any additional refining of the detail needed can literally be added in the process, this should always provide a benefit no matter what camera or megapixels you have (up to a percentage of the original of course). If you are losing detail when upscaling your print you need to look at a different software package, and of course make sure you have done everything you can to capture the detail to begin with.