Women diagnosed with cervical cancer through a smear test “have a far better chance of being cured than women who do not go for tests,” BBC News has today reported.
The news is based on Swedish research that looked at 1,230 women diagnosed with cervical cancer, examining patterns between how their disease was detected and how likely it was they would be cured and survive. Following them for an average of 8.5 years after diagnosis, it found that the cure rate was 92% among those whose cancer was detected through cervical screening and 66% among those who were diagnosed after they developed symptoms. Of note, they found a lower chance of being cured among women with symptoms who were overdue for screening.
These findings are perhaps unsurprising, as women who have developed cancer symptoms generally would be expected to have a more advanced stage of cancer than women whose cancer is detected at screening and is not yet causing them symptoms. As such, women identified through symptoms, rather than screening, may be expected to have a lower chance of being cured. The study’s results support the value of the UK’s current cervical screening programme and the importance of attending screening.
The study was carried out by researchers from Uppsala University, the County Council of Gävleborg and other institutions in Sweden. Funding was provided by grants from the Swedish Cancer Society, the Swedish Foundation for Strategic Research, the Gävle Cancer Fund, and the Centre for Research and Development, Uppsala University and the County Council of Gävleborg. The study was published in the peer-reviewed British Medical Journal.
News coverage has reflected the findings of this research.
This was a nationwide population-based cohort study looking at whether detection of cervical cancer through screening improves cancer cure and survival rates. Cure rates are of particular interest as it has been suggested that cervical screening may have the apparent effect of prolonging survival times simply because the cancer is detected at an earlier stage than it otherwise would have been (i.e. screening could cause women to just live for longer with a diagnosis of cancer). If screening actually improves cure rates this would be an important finding (though arguably this could still be just because being diagnosed at an earlier stage the cancer is more likely to be curable).
Using a cohort study to answer this question has some limitations, as the outcomes in a cohort study may be influenced by other health and lifestyle differences between those women who chose to attend screening and those who did not. These differences may be the cause of any relationship seen, meaning in this case we cannot be certain that screening is the only factor affecting survival rates.
Ideally this sort of question would be addressed using a randomised controlled trial that randomised people into different screening practices and then followed them up over time looking at cancer outcomes and cure rates. However, as cervical screening is already offered in countries such as Sweden and the UK, carrying out a randomised trial that withheld cervical screening would not be considered ethical.
The Swedish cervical screening programme invites women for screening every three years among those aged 23-50, and every five years for women aged 51-60. In the UK it is every three years between 25 and 49, and every five years between 50 and 64.
The current study linked all women with cervical cancer in Sweden diagnosed between 1999 and 2001 to the national Swedish causes of death register. The researchers then followed the women to the end of 2006 to check survival in the years following diagnosis.
The researchers analysed women separately according to their age at diagnosis (23-65 years old), including those with a diagnosis more than five years beyond the last invitation to screening (66 years or above). Screening-detected cancers were defined as cancers in women who had an abnormal smear test result recorded between one and six months prior to their diagnosis. The remaining women who did not have an abnormal smear test between one and six months prior to their diagnosis were classed as having had a ‘symptomatic diagnosis’, i.e. a diagnosis based on detectable symptoms rather than screening. Abnormal smear tests taken within one month of diagnosis were also not considered to be screen-detected, as it was considered this might have been part of the diagnostic assessment in women with symptoms of cancer.
The researchers also looked at women with symptomatic cancer who were diagnosed more than six months after their last smear test and outside of the recommended screening interval of 3.5 years if they were under the age of 54; or an interval of 5.5 years if they were 55 or over. These women were considered to be overdue for having their screening test and were compared with women who were not overdue their screening test when they were diagnosed symptomatically.
The outcomes examined were survival rates (survival in the cohort compared with expected survival in the general female population); and ‘statistical cure’ rates (defined as the women no longer experiencing any greater risk of death compared with the general female population).
This cohort of 1,230 women was followed for an average of 8.5 years after diagnosis of cervical cancer. Five years after their diagnoses 440 of the women had died, 373 of these deaths were recorded as being due to cervical cancer (31 died from other cancers, and 36 from a non-cancer cause).
The proportion for women with screen-detected cancer who survived for at least five years was 95% (95% confidence interval [CI] 92 to 97%), whereas for women with symptomatic cancers it was 69% (95% CI 65 to 73%). The cure rate for screen-detected cancers was 92% (95% CI 75 to 98%) compared with 66% (95% CI 62 to 70%) for symptomatic cancers. This 26% difference in cure rate was statistically significant.
Among women with symptomatic cancers, the proportion cured was significantly lower among those overdue for screening compared to those who had been last screened within the recommended interval (difference in cure 14%, 95% CI 6 to 23%).
Cure proportions were related to the stage of the cancer at the time of diagnosis, but even after taking into account stage at diagnosis, cure rates still remained higher among screen-detected cancers than symptomatic cancers.
The researchers conclude that screening is associated with improved rates of curing cervical cancer. They note that they cannot rule out the possibility that factors other than screening may have contributed to the differences observed. They also said that using cure as an outcome removes the problem of ‘lead time bias’ that occurs when looking at length of survival as an outcome of screening (discussed in the conclusion section below).
They recommend that further evaluations of cervical screening programmes should consider using a similar approach of looking at the proportions of women with cancer who are cured.
As the researchers discuss, women with cervical cancers detected by screening are known to have an improved chance of surviving their cancer. The study’s apparent improvement in survival outcome may be partly due to a phenomenon known as ‘lead time bias’, meaning that women diagnosed through screening are simply diagnosed at an earlier stage than they would have been if they waited for symptoms to develop. That is to say, that they might not live any longer, just live for longer knowing they had cancer, having detected it at a point before outward symptoms appear. This cohort study aimed to see whether screening improves cure rates, which the researchers hoped would avoid this problem.
A cohort study isn’t the best type of study design to assess the effect of a screening or therapeutic practice against disease outcome, as in a cohort there may be other health and lifestyle differences between women who chose to attend screening or not. The researchers themselves acknowledge that the possibility of such confounding cannot be ruled out. A more reliable way to assess this question would be a randomised controlled trial that randomly assigned women different screening practices and then followed them up over time looking at cancer outcomes and cure rates. However, as cervical screening is already offered in countries such as Sweden and the UK, blocking women access to cervical screening would not be considered ethical, and such a study is highly unlikely to be approved.
These findings are perhaps unsurprising. Women who have developed cancer symptoms are likely to have a more advanced stage of cancer than women whose cancer was detected incidentally through screening. As such, symptomatic women may have a lower chance of cure than women detected at an earlier stage. The fact that there was a lower chance of cure among symptomatic women who were overdue for screening further supports this.
However, the researchers’ further analyses suggested that this was not simply a case of the cancers being diagnosed at an early stage: though cure rate was related to cancer stage, taking into account stage at diagnosis did not remove the difference in cure rates between screen-detected and symptomatic-detected women. The reasons for this cannot be explained by this study, and as the researchers conclude, further evaluations of the benefit of cervical screening programmes should consider looking at cure proportions.
The UK has a slightly different schedule for cervical screening than Sweden, where this study was carried out. The Swedish cervical screening programme invites women for screening every three years among those aged 23-50, and every five years for women aged 51-60, while in the UK it is three-yearly between 25 and 49, and five-yearly between 50 and 64. This and other differences between the countries may mean that the results may not be representative of the UK. However, they generally appear to support the value of cervical screening programmes and the importance of women attending such screenings.
Analysis by Bazian