There is a known issue with comma seperated values and examine/lucene. The work around we use is to hook into Examine's GatheringNodeData event and split the csv by spaces rather than with commas:

using System;
using System.Text;
using System.Xml.Linq;
using Examine;
using umbraco.BusinessLogic;
using umbraco.MacroEngines;
using umbraco.presentation.nodeFactory;
using Umbraco_Site_Extensions.automation;
using UmbracoExamine;

namespace Umbraco_Site_Extensions.examineExtensions
{
    ///
    /// handle any lucene data injection here
    ///
    public class ExamineEvents:ApplicationBase
    {

        public ExamineEvents()
        {
            ExamineManager.Instance.IndexProviderCollection[INDEX_NAME].GatheringNodeData += ExamineEvents_GatheringNodeData;
        }

        void ExamineEvents_GatheringNodeData(object sender, IndexingNodeDataEventArgs e)
        {
            if (e.IndexType == IndexTypes.Content)
            {
                var node = new Node(e.NodeId);                
                InjectNewsCategoriesWithOutComma(e, node);                
            }
        }

        ///
        /// cannot do searching with comma in values properly
        ///
        ///
        ///
        private void InjectNewsCategoriesWithOutComma(IndexingNodeDataEventArgs e, Node node)
        {
            if (node.NodeTypeAlias == Constants.NewsItemTypeAlias &&
               node.GetPropertyValue(Constants.NewsCategoryFieldAlias) != string.Empty)
            {
                e.Fields.Add( Constants.NewsCategoryFieldAlias +"Searchable", node.GetPropertyValue(Constants.NewsCategoryFieldAlias).Replace(","," "));
            }
        }
    }
}

You'll also need to add the new field into the IndexUserFields element as well but otherwise you should be good to go