Cancer risk, premature death and sleeping tablets?

@HealthMed A team of researchers from Scripps Institute and the Jackson Hole Center for Preventive Medicine in Jackson, Wyoming, in the United States has reported in a publication in the British Medical Open Journal that use of sleeping tablets is associated with increased risk of cancer (lymphoma or cancer of the lung, colon or prostate), and  premature death. Results were obtained by interrogating electronic case records within a large US health care service: the Geisinger Health System (GHS), the largest rural integrated health system in the USA. GHS serves a 41 county area of Pennsylvania with approximately 2.5 million people.
This report has been disseminated widely in the international media, including for example in the Los Angeles Times.
The 2 major questions raised by this work are whether findings are explained by effects caused by the sleeping tablets -  or are the kind of people who are prescribed hypnotics just more likely to develop cancer or premature death, as a result, for example, of genetic factors, lifestyle reasons or associated disease causing the sleep disturbance.
Obvious more specific questions include are whether this relationship is:
- caused by the sleeping tablets (hypnotics);
- a feature of 'reverse causation - ie resulting from the initial sleep disturbance having been caused by symptoms of at that stage undiagnosed cancer, or sleep disturbance caused by cancer risk factors such as smoking, excess alcohol or obesity - ie some cancer-related factor leading to patients seeking a prescription for hyponotics
- coincidental
- or spurious - the result of some bias in subject selection.
The short, ~2.5 year average treatment duration makes this possible early cancer effect important to exclude as an explanation.
The authors say they have made reasonable efforts to exclude other causes of increased cancer risk. A wide range of hypnotics were included in the study, making it unlikely that a direct chemical action of the tablets is responsible for the cancer risk.
An intriguing possibility is that the association is a marker of chemical changes in the brain arising from sleep disturbance: it is well recognized that disordered brain function leads to production by the brain of inflammatory mediators (cytokines). One consequence is the thought to be the increased risk of heart disease reported in patients who have depression. That has lead to interest in using new biological agents to inhibit the action of these cytokines as a new approach to treatment depression and other serious neurological disorders. Taking sleeping tablets, although prolonging sleep duration, may not be sufficient to induce a normal sleep pattern.
In the context of the Scripps study, this raises the question whether  inflammatory mediators
could modulate tumour formation or accelerate the growth of early stage cancers. This concept was discussed by Reiche and colleagues in a Lancet Oncology review in 2004. As an example, the chemical - tumour necrosis factor alpha - is a potent cytokine implicated in depression and - as the name indicates - with a role in tumour biology, for example in modulating tumour blood supply.
Although in the Scripps study the reported relative risk of cancer increased from 3.5 to 5-fold with the numbers of sleeping tablets used, there needs to be a clear description of absolute events rates. And the increase in premature death was from a very wide range for causes, from accidents to asthma, flu and diabetes.
Of note, this was an observational study controlled by the authors by aiming to find well-matched subjects not on sleeping tablets (cases: 10 529 patients who received hypnotic prescriptions;  23 676 matched controls). A randomised controlled trial aimed at assessing these particular outcomes would have been needed to test these findings with more confidence of lack of bias. 
If the relationship is confirmed as having a causative link, the study would illustrate yet again the importance of balancing benefit and risk when prescribing tablets, and ensuring that any prescription is  continued for the minimum clinically indicated time.