In 2016 the average time between symptom onset and starting treatment for pulmonary TB in England was 77 days(2). A key aim of Public Health England and NHS England’s Collaborative Tuberculosis Strategy is to “improve access to services and ensure early diagnosis” (20). This study has identified that, since the implementation of the CRP, migrants in East London are more likely to experience delay in treatment of their TB. This is an important finding since the longer the delay in diagnosing TB, the greater the morbidity(21) and mortality(22) for the individual, and the greater the risk of transmission in the community(23).
There are a number of potential mechanisms at play which may be relevant in the association between the experiences of migrants accessing healthcare and health policies that restrict access to care based on immigration status(11,12). There is already evidence to show that migrants in the UK are often not aware of their entitlements to care(24). This is further complicated by recent legislative changes that have altered who is eligible for care and who is not, including the introduction of an immigration health surcharge that accompanies visa applications(25). Previous research has shown patients’ concerns about being charged for care delay health-seeking, even before a diagnosis has been made(26–28). This is important because diagnosis and treatment of TB, like many other infectious diseases, is exempt from charging in the UK, regardless of immigration status(29). Crucially, however, patients present with undifferentiated symptoms - not a diagnosis - and many may be unaware of the details of the regulations(30).
Recent evidence suggests the CRP is just one policy among several which constitute a broader ‘hostile environment’ aimed at people living in the UK illegally(31). For example a data-sharing agreement between NHS Digital and the Home Office(32,33) requires that NHS staff report people with an outstanding bill exceeding £500 to the Home Office. Significant sharing of patients’ demographic information between the NHS and the Home Office has been reported(34).
The ‘hostile environment’ is an important socio-political context within which people make decisions about seeking help, including decisions regarding whether and how they identify themselves as a candidate for health care(35). This effect may be independent of their legal eligibility for free care. Fear as a deterrent to healthcare access among migrants has been well documented in several countries including the UK(24,36). In 2012 the government made clear their explicit intention to create a “really hostile environment” for those living in the UK without the legal right to do so(37). Aside from the CRP, other measures included: a rise in immigration raids; the prospect of unlimited detention; the threat to health and even life that immigration detention poses(38,39); school meals withheld because of parents’ immigration status(40), restrictions to the housing rental market, driving licenses and bank accounts in similar ways to the restrictions applied to the NHS(41). The influence of these other ‘hostile environment’ policies have not been accounted for within this study but may have contributed to the overall findings.
There are other contemporaneous policy-related contexts which may be relevant to the increase in time to treatment reported in this study. Access to translation services have changed due to imposed cuts under conditions of austerity. Across the NHS, staffing levels decreased and waiting times for hospital care increased during the study period amidst concerns of an NHS ‘in crisis’(42). It is important to consider whether local changes in service provision may have impacted the results. The study data is collected from three London boroughs each served by a hospital with an A&E and local respiratory and TB services. Whilst there was an increased focus on TB among migrants during the study period, including education and awareness raising in primary care and local communities, no other major service changes occurred during this time. Finally, as with all observational analyses, causal associations cannot be inferred between the implementation of the CRP and the significant increase in the time to diagnosis.
Our study shows the UK-born population experienced a non-significant increase in the time to treatment of TB. A significantly higher proportion of UK-born patients had one or more social risk factors compared with those not born in the UK. This is reflective of national data(2). Therefore the UK-born population is potentially more vulnerable to the effects of austerity; health suffers whilst individuals manage other competing priorities such as employment, housing, and limited income (earned or through welfare) restricting their means to access care(43). This would result in an underestimation of effect. Conversely, clinicians’ sensitivity to TB as a differential diagnosis is likely to be heightened among patients with particular risk factors such as homelessness. However, not being born in the UK – particularly individuals from high TB incidence countries – is also likely to increase alertness amongst clinicians to the possibility of a TB diagnosis.
Nevertheless, these factors do not explain why only the non-UK born population experienced a significant delay in time to diagnosis and treatment following the introduction of the CRP whilst the UK-born population did not. Of note, migrants diagnosed with TB after 2014 had been in the UK significantly longer than those diagnosed before. There may be a number of reasons for this such as changes to immigration policies and the introduction of pre-entry TB screening(44). Nonetheless, ‘newness’ of migrants has been associated with increased difficulty accessing care(35,45) potentially resulting in an under estimation of the effect size.
Whilst the study area is geographically small, it accounts for approximately 10% of all cases of TB in the UK during the research period (2). TB in the UK is largely focused in urban areas with large migrant communities. Thus it is possible our findings may be applicable to other areas within England which have similar migrant populations and have been subject to similar policies designed to restrict access to healthcare for some migrants and visitors. However, this study does not demonstrate causality and further research is required to examine the nature of the relationship between different categories of migrant, their eligibility for free NHS care and the complex and evolving arena of laws, policies and practices which shape access to TB treatment.
There are a number of limitations to the study. This paper does not claim causality but nevertheless demonstrates an important association that warrants further investigation. The ability to communicate in the language of the host country has been shown to affect healthcare access as well as the quality of care received by migrants(46), however English language ability is not routinely collected. Immigration status has also been shown to affect healthcare access(47) but is not routinely collected. It was therefore not possible to differentiate between migrants eligible for free NHS care and those who are not. Socio-economic status was determined through a proxy measure – occupational status – which has well-documented limitations(48). A binary cut-off before and after the CRP does not reflect the reality of a policy which was rolled out over subsequent years. The unbalanced nature of the numbers in the UK-born and non-UK-born groups should not have introduced bias to the tests for significance used here but could have had an impact on the power of these tests to detect significant differences. One potential way to try to address this asymmetry in future work could be to employ matching, however careful thought would have to go into the selection of matching criteria. There are other techniques that could have been used to analyse the data that would account for the longitudinal nature the dataset, such as an interrupted time series analysis. However, this would introduce additional assumptions such as linearity of the data, predictable change in time-varying external factors, and autocorrelation. The findings presented here merely suggest an association with time as a binary variable using a simple test of proportion.