Presuming you are using MS SQL Server there are two approaches to achive what you are looking for:
1) Use trigger as suggested by other users. Trigger would be for INSERT / UPDATE that will check the date ranges and will allow new values to be added or raise error.
2) You can use composite primary key in ItemTable:
CREATE TABLE [dbo].[ItemTable](
[Id] [int] IDENTITY(1,1) NOT NULL,
[HeaderRef] [int] NOT NULL,
[SLRef] [int] NOT NULL,
PRIMARY KEY CLUSTERED
(
[HeaderRef] ASC,
[SLRef] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
) ON [PRIMARY]
GO
Now this will put a constraint on the ItemTable and sql server WILL NOT allow duplicate combination of headerRef and SLRef int values (keys).
Back you your HeaderTable, you can put the unique constraint to stop duplicating the range of start and end dates
CREATE TABLE [dbo].[HeaderTable](
[ID] [int] IDENTITY(1,1) NOT NULL,
[STARTDATE] [datetime] not NULL,
[ENDDATE] [datetime] not NULL,
CONSTRAINT [PK_HeaderTable] PRIMARY KEY CLUSTERED
(
[ID] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
) ON [PRIMARY]
GO
Now create unique index on HeaderTable for start and end dates.
/****** Object: Index [IX_HeaderTable] Script Date: 03/13/2017 12:24:51 ******/
CREATE UNIQUE NONCLUSTERED INDEX [IX_HeaderTable] ON [dbo].[HeaderTable]
(
[ENDDATE] ASC,
[STARTDATE] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, SORT_IN_TEMPDB = OFF, IGNORE_DUP_KEY = OFF, DROP_EXISTING = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
GO
You can also put another constraint on HeaderTable that will check if start date cant be before end date.
ALTER TABLE [dbo].[HeaderTable] WITH CHECK ADD CONSTRAINT [CheckEndLaterThanStart] CHECK (([ENDDATE]>=[STARTDATE]))
GO
ALTER TABLE [dbo].[HeaderTable] CHECK CONSTRAINT [CheckEndLaterThanStart]
GO
Hope this helps!