From December 1, 2014, to December 31, 2015, we continually enrolled 252 culture-confirmed pulmonary TB patients diagnosed in the Songjiang district, Shanghai. Of the 252 cases enrolled, four (1.6%) were excluded: three patients had an infection with non-tuberculosis mycobacteria (NTM), and one other case died of respiratory failure during the anti-TB treatment. Among the remaining 248 participants whose data were included for analyses, 29.8% (74 of 248) were residents, and 70.2% (174 of 248) were internal migrants. The characteristics of the study population were detailed in Table 1.
Internal migrant TB patients were significantly younger than resident cases (median age, 32 years versus 52 years; p<0.01), had a higher annual household income (median 90 000 versus 70 000 Chinese Yuan [RMB], p=0.01). However, migrants with TB had a significantly smaller living space per capita compared to residents (12 [IQR, 8-20] vs. 34 [IQR, 24-60] square meters, p<0.01). Due to restrictions on medical insurance among internal migrants, half (52%, 90 of 174) of the migrant cases paid all of the medical costs out-of-pocket by themselves. In contrast, most (95%, 70 of 74, p<0.01) residents had health system insurance that covered the related medical expenditures. Also, internal migrants had a significantly lower rate of hospitalizations than that of resident patients during both the periods before and after the diagnosis of TB (24%, 42 of 174 versus 58%, 43 of 74, p<0.01).
Most of the TB patients during the study period were either cured (34%, 85 of 248) or completed the treatment (64%, 158 of 248), and such proportions were similar between internal migrant and resident TB patients. Patients went to the hospital for an average of three times before being diagnosed with TB and went to the hospital 14 more times during TB treatment. Patients delayed an average of 10 [IQR 3-22] days to seek medical care after well-defined TB symptoms occurred, and the hospital delayed an additional 13 [IQR 6-20] days to make the diagnosis.
Direct costs and indirect costs of TB services
The direct costs and lost income are summarized in Figure 1. In general, residents paid almost twice as much as internal migrants for direct costs and lost income (25 824 and 13 816 Chinese Yuan [RMB], respectively). The majority of the total expenses (> 65%) were direct medical costs. The proportion of total costs that were direct medical costs was similar among residents and internal migrants (67.6% vs. 69.0%). Non-medical items, including transportation, extra food supplies, accommodation, and guardian care, accounted for a higher proportion of total costs among residents compared to internal migrants (8.1% [95%CI 7.8%-8.5%] vs. 6.2% [95%CI 5.8%-6.6%], p<0.01). The other quarter of total costs to TB patients were from indirect income loss. The proportion of indirect costs among the total costs did not differ significantly between residents and internal migrants (24.3% vs. 24.8%, p=0.30).
Costs incurred among different treatment stages.
The costs accrued before getting a TB diagnosis were one-third of the total expenses, and the proportion of costs accrued before getting a TB diagnosis was slightly higher among residents compared to internal migrants (35.4% [95%CI 34.8-36.0%] vs. 31.9% [95%CI, 31.1-32.6%], p<0.01). Lost income increased, and medical costs decreased slightly after diagnosis; however, the costs of medicine after diagnosis rose from 28.1% to 44.7% of the total expenses among residents and from 31.7% to 47.2% among internal migrants.
Catastrophic costs and risk factors
Overall, 55 (22%, 55 of 248) TB patients have experienced household catastrophic costs during the entire illness, and more than half of them had experienced high TB care-related expenses (> 10% of total household income). After adjusting for the migration status in the multivariate model, TB patients who experienced hospitalization (adjusted OR [aOR], 10.08, 95%CI [4.96-23.41]), having no available health insurance (aOR, 2.69, 95%CI [1.68-6.72]), and being older than 45 years (aOR, 2.50, 95%CI [1.20-5.21]) were independently associated with the occurrence of catastrophic costs (Table 2). The benefits of reimbursement for TB diagnosis and treatment had less impact on the prevalence of the catastrophic cost by reducing it from 22% to 20% (49 of 248, p=0.50). Besides, TB patients who experienced hospitalization and were older than 45 years remained significant for catastrophic costs after this specific reimbursements, except the health insurance (Table 2).
Migrant TB patients had higher income levels and were less likely to be hospitalized than residents who were TB patients. The proportions of household’s catastrophic costs were significantly different between households of residents versus household of internal migrants (p<0.01; Table 2). However, after adjustment for hospitalization, the health insurance, and age groups, there was no significant difference in the prevalence of catastrophic costs between migrants and residents (adjusted OR, 0.64, 95%CI [0.25-1.61]). The average proportion of the total costs that were covered by insurance payments was 18.8% (95%CI 14.0%-23.7%) among internal migrants compared to 54.0% (95%CI 48.5%-59.4%) of residents.
Hospitalization and health insurance among migrants and residents
Hospitalization rates were significantly higher among resident patients (58%, 43 of 74) compared to migrant patients (24%, 42 of 174, p<0.01), and such a difference mainly existed in a patient group of individuals less than 45 years old (p<0.01, Figure 2). Migrant patients had a significant trend association between hospitalization and age (p<0.01), while resident TB patients had a high average rate without the trend association (p=0.59, Figure 2). In total, patients without health insurance were less likely to be hospitalized than those with health insurance (24%, 23 of 94 vs. 40%, 62 of 154, p=0.01) (Figure 2). Since only 5% of the resident patients lacked health insurance, we stratified this analysis by migration status; however, we observed no significant factors for the prevalence of hospitalization in either resident or migrant patients subgroups. In a multivariable logistic regression model, TB patients who were a migrant (aORs of 0.29 [95%CI 0.14-0.56]) and age between 25-34 years old (0.28 [95%CI 0.12-0.67]) were less likely to be hospitalized.
We also stratified the hospitalization by pre-diagnosis and post-diagnosis: 79% (67 of 85) patients had hospitalization before the diagnosis of TB, and 13% (11 of 85) of them had hospitalization during both periods. Among migrants, more TB patients (83%, 35 of 42) had a pre-diagnosis hospitalization compared to local residents (60%, 32 of 53, p=0.01).