I did search on b144 for both b38 and b37. Neither rsid and nor any variant at that reported position is present in both the VCFs. You seem to be correct and you may have to contact NCBI/dbSNP maintainers regarding this discrepancy.

On July 17th 2015 I observed a similar issue that I reported to NCBI Helpdesk: "I am wondering why the following pathogenic ALK variant http://www.ncbi.nlm.nih.gov/clinvar/variation/18083/#clinical-assertions is not present in the VCF file downloaded from here (downloaded July 17th, 2015): ftp://ftp.ncbi.nlm.nih.gov/pub/clinvar/vcf_GRCh37/clinvar_20150629.vcf.gz" This was the reply: "Thanks for reporting this to us. I have confirmed your report. Doubling checking this against the All_#.vcf.gz file for b144 mapped to GRCh37 and 38, I also see that this rsID is in those two vcf files. I will check with our developers to ask them looking into this issue.Thanks for reporting this to us." So there are definitely SNPs missing from some VCF files. I have not heard back from NCBI since, so I don't know if the problem persists.

I'm having the same issue. In fact I did a count of all the SNPs on chromosome 22 in the dbSNP BED format files and then did the same for the VCF file and found that there are +/-50 000 SNPs missing from the VCF file on chromosome 22 alone. @Christian Did you ever hear back from the NCBI help desk?