Clinical and economic impact of a specific BCG vaccination program implemented in Prato, central Italy, involving foreign newborns on hospitalizations

ABSTRACT

In Tuscany (Central Italy), the average annual notification rate of tuberculosis (TB) in the years 2007–2012 was 7.5–9.8 per 100,000 people, with the Local Health Unit of Prato (LHU4) showing the highest rate compared to the other regional area. Therefore, in order to reduce the burden of TB, foreign newborns in the LHU4 are being given the Bacillus Calmette-Guérin (BCG) vaccine since 2000. The aim of this study is to assess the impact of BCG vaccination in Prato, in terms of TB-related hospitalizations and costs.

The regional archive containing all TB-related discharges and costs in the period 2007–2014 was consulted. Data regarding foreigners living in the LHU4 who have been vaccinated since 2000 were compared with those living in the other Tuscan LHUs and never vaccinated. These populations were then disaggregated by a threshold age of 15 y.

After calculating the standardized hospitalization rates, the expected number of hospitalizations for TB among unvaccinated adults (in both populations) was found to be similar in the LHU4 and the other LHUs (165 vs. 156). However, expected number of hospitalizations among children in the other Tuscan LHUs (67) was double that of the LHU4 (34). If the same vaccine had been administrated everywhere, each year 29 hospitalizations could have been avoided and EUR 343,525 saved. Overall, BCG vaccinations cost EUR 14,879 in the LHU4, but 69 hospitalizations were avoided and EUR 107,435 saved.

The introduction of the BCG immunization program in the LHU4 of Prato has led to significant reductions in the clinical and economic impact of TB.

KEYWORDS:

• BCG
• economic evaluation
• foreigners
• hospitalizations
• infectious disease
• vaccination

Introduction

Tuberculosis (TB) continues to be a relevant public health problem today; it has a great impact worldwide, especially in low-income countries. Nine million new cases are notified every year. There are around 50 million people suffering from TB in the world today and 2 million subjects die every year because of it. In particular, immigration, the spread of AIDS, and the poverty in the suburbs of metropolises have contributed to the prevalence of this disease in high-income countries as well.¹

During 2014 incidence rate was < 10/100,000 in middle and northern Europe (in Germany it was 6.2, in France 8.7, in Austria 7.8, in Sweden 7.5, in Switzerland 6.3/100,000); in Spain and United Kingdom it was 12/100,000, in Portugal 25 and much higher in eastern Europe.^2,3 In Italy, 3,142 cases of TB (68% of which were pulmonary TB) were notified during 2012, with an incidence rate of 5.2/100,000 inhabitants. Of these cases, 58% involved foreigners.^4,5 In 2013, there were 3,153 notified cases in Italy with an incidence rate of 5.7/100,000 inhabitants. About 5% of the newly diagnosed people were younger than 15 y of age. The number of diagnosed males was 3 times higher than that of females.⁶

In Tuscany, the average annual notification rate of TB in the period 2007–2012 ranged from 7.47 to 9.76 cases per 100,000 people. In the Local Health Unit (LHU) 4 in Prato, the rate was 2 to 3 times higher compared to the other Tuscan LHUs every year.⁷ The larger number of TB cases can be explained by the higher percentage of foreigners living in Prato (14.7%), compared to the percentage in the whole of Tuscany (10.3%) and Italy (8.1%) in 2014.⁸

At present, the only vaccine available to prevent TB is the Bacillus Calmetteand Guérin (BCG) attenuated vaccine.⁹ Several reviews, meta-analyses, and case studies have tried to quantify the protective efficacy of BCG, with very heterogeneous results. The causes for variable efficacy have been discussed in a WHO document on BCG vaccine.¹⁰ One of them is the genetic variation in the used BCG strains (BCGs Japan, Birkhaug, Sweden, and Russia and the late strains, including BCGs Pasteur, Danish, Glaxo, and Prague), related to genetic polymorphisms produced during different methods of vaccine preparation.¹¹ This variability depends also on the disease outcome being measured, on the methods and routes of vaccine administration and on the environment and characteristics of the studied population (genetic population characteristics,¹² interference by no-tuberculosis mycobacteria It results in a no-specific immune response against mycobacteria, whom vaccine does not add anything,^13,14,15 or interference by concurrent parasitic infection, which makes vaccine response less effective¹⁶). The efficacy for consistency of different strains ranges from 0% to 80%,¹⁷ while the overall protective effect of BCG vaccine, as average, has been found to be 51% in the clinical trials (95% CI, 30–66%) and 50% in the case-control studies (95% CI, 36–61%) according to a review,¹⁸ using a model that included the geographical latitude of the study site and the data validity score as covariates. In a more recent case-control study, the efficacy for consistency against TBC symptomatic disease was estimated as average at 50% (95% CI, 0.39–0.64).¹⁹ Other trials showed a efficacy for consistency of 71% (RR, 0.29; 95% CI, 0.16–0.53) with regard to TB-related deaths, and of 64% (OR, 0.36; 95% CI, 0.18–0.70) from TB-induced meningitis.¹⁹ Another recent review revealed a protection rate of 19–27% against infection and 71% against active TB.²⁰

In newborns and children, the estimated protective efficacy rate was found to range from 52% to 100% in terms of prevention of tuberculous meningitis and miliary TB and from 2% to 80% for the prevention of pulmonary TB.

In Italy, the indications of the BCG vaccine include the prevention of TB infection, strengthening of the individual's immune system against the Mycobacterium, and the improvement of antibody protection against the development of the disease.⁹

Decree number 465 from the President of the Italian Republic, signed on 11 July, 2001, recommended BCG vaccination for newborns and children younger than 5, with negative tuberculin tests for those living closely with people affected by contagious TB. The vaccination was not recommended for children aged 10–14 y. The decree also suggested that the immunization be evaluated for the other subjects (aged 5–10 years), eventually with a tuberculin test. In addition, the Italian Minister of Health recommended²¹ offering this vaccine at the LHU level to newborns and children younger than 5, who have been in a TB-endemic country for more than 6 months, are living in a community with a higher risk of TB, or are coming from countries where this vaccine is offered.

However, because of the high percentage of foreigners living in Prato, a specific BCG vaccination program was implemented in the LHU4 in Prato long before this law. As a matter of fact, since 2000, this vaccine has been offered for newborns with at least one foreign parent whose native country is TB-endemic.

Therefore, the aim of the current study is to assess the impact of BCG vaccination on foreign newborns in the LHU4 in Prato in terms of TB-related hospitalization rate and costs, compared to all other areas in Tuscany where the BCG vaccine has been not offered by LHUs.

Results

In terms of hospitalizations, in the LHU4 in Prato, the most frequently hospitalized age group was 5–9 y (14 hospitalizations from a total of 35). Among adults (15 y or older), the most frequently hospitalized age groups were 35–39 y (27 hospitalizations/127) and 30–34 y (23/127). From the age of 39 years, the number of hospitalizations reduced steadily with increase in age, and there was just one hospitalization in the age group 65–69 y.

Among foreigners aged less than 15 y living in the purview of other Tuscan LHUs, the number of hospitalization reduced with age (170/468 in the age group 0–4 years, 169/468 in the age group 5–9 years, and 129/468 in the age group 10–14 years). Among adults (15 y or older), the pattern resembles that of the LHU4 in Prato: they were more often middle-aged and the 2 age groups most represented were 30–34 y (208/1126) and 25–29 y (202/1126). From the age of 34 years, the number of hospitalizations reduced regularly and only one case was found among those older than 85.

The TB-related hospitalization rates in foreigners both younger and not younger than 15 y of age, living in the purview of the LHU4 in Prato and other Tuscan LHUs, are shown in Table 1. Both absolute number of hospitalizations and crude rates peaked in the LHU4 in Prato in 2008 and 2012 and in the other LHUs in the period 2011–2013, for subjects younger than 15 y. The highest hospitalization rate in the LHU4 in Prato was in 2008 (107.0 hospitalizations/100,000 inhabitants) and in the other LHUs in 2011 (193.7 hospitalizations/100,000 inhabitants). The maximum number of hospitalizations in adults (not younger than 15 y of age) was reported between 2008 and 2010 in both the LHU4 in Prato and the other LHUs. The hospitalization rate after 2010 showed a slight but steady decreasing trend in adult populations in both the LHU4 in Prato and the other LHUs, but was on the whole stable. In the analysis period of 8 years, the crude average hospitalization rate in foreign children living in the purview of the LHU4 in Prato (58.3/100,000 inhabitants) was about half of the corresponding rate in Tuscany (109.4/100,000 inhabitants). Meanwhile, the crude average hospitalization rate in the older subjects was comparable in the LHU4 in Prato and the other LHUs (28.7 and 27.3 per 100,000 inhabitants respectively).

Table 1. Number of hospitalizations and hospitalization rates related to TB by age groups in the LHU4 in Prato and in the other Tuscan LHUs.

Hospitalizations due to TBC in Tuscany, Italy
	Foreigners living in the LHU4 in Prato < 15 years		Foreigners living in the LHU4 in Prato ≥ 15 years		Foreigners not living in the LHU4 in Prato < 15 years		Foreigners not living in the LHU4 in Prato ≥ 15 years
	Number of hospitalizations	Hospitalization rate/100,000 inhabitants	Number of hospitalizations	Hospitalization rate/100,000 inhabitants	Number of hospitalizations	Hospitalization rate/100,000	Number of hospitalizations	Hospitalization rate/100,000 inhabitants
2007	2	34.0	12	25.9	13	34.5	131	34.6
2008	7	107.0	21	42.5	38	86.6	162	36.0
2009	4	57.9	21	41.1	27	54.6	141	27.5
2010	1	13.4	18	32.5	50	92.7	154	27.5
2011	6	75.0	14	23.4	111	193.7	131	21.7
2012	7	89.2	17	29.2	99	192.0	151	28.6
2013	6	68.9	9	14.0	73	129.3	121	21.2
2014	2	21.1	15	21.3	57	91.6	135	21.3
Total	35		127.0		468.0		1,126.0
Average	4.4	58.3	15.9	28.7	58.5	109.4	140.8	27.3

In the LHU4 in Prato, the average hospitalization rates per 100,000 inhabitants in the 2 age groups (< 15 y and ≥ 15 y) were respectively 0 and 28.13 for other forms of respiratory TB, 315.68 and 14.06 for primary TB infection, 26.31 and 12.31 for TB of other organs, 0 and 3.51 for genitourinary TB, 0 and 3.52 for meningeal or central nervous system TB,0 and 1.76 for intestine-peritoneum-mesenteric gland TB, 0 and 1.76 for miliary TB, and 105.22 and 161.73 for pulmonary TB.

In other Tuscan LHUs, the average hospitalization rate per 100,000 inhabitants in the 2 age groups (< 15 y and ≥ 15 y) were respectively 5.82 and 16.24 for other forms of respiratory TB, 638.16 and 9.82 for primary TB infection,13.58 and 22.65 for TB of other organs, 0 and 0.76 for genitourinary TB, 3.88 and 3.96 for meningeal or central nervous system TB, 0 and 6.23 for intestine-peritoneum-mesenteric gland TB, 1.94 and 2.45 for miliary TB, and 223.06 and 148.57 for pulmonary TB.

The average length of stay showed similar values among foreigners younger than 15 y in the LHU4 in Prato (5.8 days) compared to children in other LHUs (5.2 days). Among adults, it was 16.7 d in the LHU4 in Prato and 23.2 in the other LHUs.

Total TB-related costs (sum of domiciliary therapy and hospitalization) depend on the number of hospitalization cases and the average cost per case. This last cost differs significantly for different age groups (Table 2). The average cost of hospitalization in subjects less than 15 y of age was EUR 1,411 in the LHU4 in Prato and EUR 1,339 in the other Tuscan LHUs, respectively. Average costs for adults hospitalized for TB were higher (EUR 5,021) in the LHU4 in Prato than in the other LHUs (EUR 4,519).

Table 2. Total costs and average costs for hospitalization in Tuscany, Italy.

Total costs and average costs for hospitalization due to TBC in Tuscany, Italy
	Foreigners living in the LHU4 in Prato < 15 years		Foreigners living in the LHU4 in Prato ≥ 15 years		Foreigners not living in the LHU4 in Prato < 15 years		Foreigners not living in the LHU4 in Prato ≥ 15 years
	Total costs	Average cost per hospitalized case	Total costs	Average cost per hospitalized case	Total costs	Average cost per hospitalized case	Total costs	Average cost per hospitalized case
2007	2,710	1,355	63,536	5,295	24,763	1,905	571,407	4,362
2008	14,939	2,134	101,168	4,818	52,662	1,386	624,812	3,857
2009	4,336	1,084	111,653	5,317	39,386	1,459	629,369	4,464
2010	1,584	1,584	89,401	4,967	69,475	1,390	675,525	4,387
2011	14,936	2,489	61,182	4,370	131,802	1,187	648,485	4,950
2012	8,672	1,239	81,486	4,793	94,942	959	740,229	4,902
2013	5,962	994	46,046	5,116	102,839	1,409	542,573	4,484
2014	813	407	82,367	5,491	58,101	1,019	640,595	4,745
Total	53,952		636,839		573,970		5,072,995
Average	6,744	1,411	79,605	5,021	71,746	1,339	634,124	4,519

Average costs disaggregated by types of diagnosis in the LHU4 in Prato and in the other LHUs were, respectively, EUR 5,285 and EUR 4,872 for other forms of respiratory TB, EUR 1,507 and EUR 1,134 for primary TB infection, EUR 2848 and EUR 2,918 for TB of other organs, EUR 0 (no cases in Prato) and EUR 3,157 for genitourinary TB, EUR 4,511 and EUR 4,722 for meningeal or central nervous system TB, EUR 3,057 and EUR 3,653 for intestine-peritoneum-mesenteric gland TB, EUR 4,054 and EUR 4,580 for miliary TB, and EUR 5,154 and EUR 4,489 for pulmonary TB.

Table 3 shows the expected number of TB-related hospitalizations in the LHU4 in Prato and in the other Tuscan LHUs, calculated by applying the direct standardization method. On average, the expected number of hospitalizations among adults was similar in the 2 populations (165 vs. 156), while expected number of hospitalizations among children was higher (approximately double) in the other LHUs (67) compared to ones in the LHU4 in Prato (34).

Table 3. Expected hospitalizations due to TB in Tuscany applying the direct standardization method.

Expected hospitalisations due to TB in Tuscany
	Expected hospitalizations in LHU4 in Prato < 15 years	Expected hospitalizations in LHU4 in Prato ≥ 15 years	Expected hospitalizations in other Tuscan LHUs < 15 years	Expected hospitalizations in other Tuscan LHUs ≥ 15 years
2007	15	110	15	147
2008	54	212	44	180
2009	33	232	31	155
2010	8	200	57	169
2011	49	155	127	144
2012	53	171	114	168
2013	45	89	84	135
2014	15	150	66	150
Total	272	1,319	538	1248
Average	34	165	67	156

These last results can be better understood with standardized hospitalization rates, as shown in Table 4. Standardized hospitalization rates and comparative hospitalization ratios show that foreigners aged at least 15 y living in the purview of the LHU4 in Prato were admitted to hospital for TB only 6% more frequently than the foreigners living in the other LHUs in Tuscany. The rates are therefore similar. On the other hand, foreign children (younger than 15 years) not living in the purview of the LHU4 in Prato were admitted to hospital almost 2 times more often than the ones living in Prato.

Table 4. Standardised hospitalization rate, comparative hospitalization ratio, hospitalizations and costs (EUR) saved in the LHU4 in Prato.

Hospitalizations and costs (EUR) of TB in Tuscany
	LHU4 Prato	Other LHUs
Standardised hospitalization rate/100,000 foreigners < 15 years	57.4	113.4
Comparative hospitalization ratio	2.0
Expected hospitalizations < 15 y in the other Tuscan LHUs with the same hospitalization rate of the LHU4 in Prato		30
Avoidable costs for the other Tuscan LHU (EUR)		343,525
Costs saved in the LHU4 in Prato (EUR)	107,435
Hospitalizations avoided in the LHU4 in Prato	69
Standardized hospitalization rate/100,000 foreigners ≥ 15 years	28.1	26.6
Comparative hospitalization ratio	1.1

If the population of children younger than 15 y living in the purview of other Tuscan LHUs had the same hospitalization rate as that of the LHU4 in Prato (57.4/100,000, instead of 113.44/100,000), there would have been, on average, just 30 hospitalizations each year instead of 59 (annual average expected number of hospitalizations calculated by applying the standardized hospitalization ratio of Prato and other Tuscan LHUs to their respective populations). Therefore, each year 29 hospitalizations could have been avoided by implementing the BGC immunization program.

Multiplying the number of avoided hospitalizations per year for the observed 8-year period (2007–2014), with the average cost per hospitalization (cost of hospitalization EUR 1,411 plus the cost of the home therapy after discharge EUR 146), a total of EUR 343,525 could have been saved with BCG vaccination (in the best scenario involving the cheapest home treatment).

In the LHU4 in Prato, with the total cost of the vaccination program amounting to EUR 14,879 in 8 y (9,591 doses and EUR 1.56 per dose)(Table 5), 69 TB-related hospitalizations were avoided in Prato, and EUR 107,435 was saved. In a more realistic scenario, with some resistant mycobacteria and the use of non-combined drugs, these savings could have been higher. In addition, these results were obtained at an average vaccination coverage of 65%.

Table 5. Administered doses of BCG vaccine and related costs (EUR) of immunization program in foreign new-borns in the LHU4 in Prato in the period 2001–2014.

BCG vaccination in the LHU 4 Prato
	BCG vaccine doses administered	Coverage	Total costs of BCG vaccination
2001	350	n.a.	546.0
2002	n.a.	n.a.	n.a.
2003	424	56.2	661.4
2004	613	86.6	956.3
2005	162	17.3	252.7
2006	669	58.4	1,043.6
2007	1,107	80.9	1,726.9
2008	1,049	70.1	1,636.4
2009	977	65.4	1,524.1
2010	1,070	71.7	1,669.2
2011	961	63.5	1,499.2
2012	964	66.9	1,503.8
2013	993	63.9	1,549.1
2014	1,054	73.2	1,644.2
TOTAL	10,141	64.5	15,820.0

n.a. = not available

Discussion

Similar studies on the clinical and economic impact of a specific BCG vaccination program involving foreign newborns were not found in literature. Therefore, the present study, even if improvable through a better data reporting, can be considered original and the results are not comparable with other studies.

Today TB can be considered a public health problem even in specific areas of developed countries, such as in the city of Prato, Tuscany. As a matter of fact, the LHU4 in Prato has always had a higher number of crude notifications of TB compared to the other LHUs in Tuscany. According to regional Tuscan database of notifications, the average notification rates of TB among foreigners in the LHU4 in Prato aged less and not less than 15 y during the period 2007–2013 were 12.73/100,000 and 71.55/100,000 respectively. In the other LHUs, these rates were 7.94/100,000 and 26.27/100,000 respectively. However, these data could be considered imprecise because, in the period under study, the average number of notifications in the LHU4 in Prato was greater for the 2 age groups (2 and 39 respectively) than the average number of hospitalizations (4 and 16 respectively). In the other LHUs, however, the rate of underreporting is more evident: the average number of notifications during the whole period for the 2 age group was 4 and 133 respectively, versus an average number of hospitalizations of 59 and 141 respectively. It is true that the TB incidence is higher in the LHU4 in Prato compared to the other Tuscan area because of the higher number of foreigners coming from endemic countries. Therefore, the absolute number of cases in the whole population is higher, as is the incidence rate. Therefore, we can assume that a cluster effect takes place because foreigners living close together often have a higher risk of transmission. However, comparing the incidence of TBC in adult foreigners in the LHU4 in Prato and the other Tuscan LHUs, the hospitalization rates are similar. Therefore, the assumption that foreigners living in Prato get sick more frequently than foreigners living elsewhere in Tuscany can be considered false.

Notification data for TB would have provided us with more precise information about the impact of tuberculosis in Tuscany, including the patients that were treated at home and not hospitalized. We did indeed ask for the total number of notifications of TB cases from the Statistical Regional Office, because TB is an infectious disease which must mandatorily be notified within 48 hours. We expected that these numbers would be similar or higher than the number of hospitalizations, because if a TB case is notified, the patient is always admitted to hospital for medical examination and the most suitable treatment. If the diagnosis is not confirmed and it is not even a mycobacteriosis, the notification is deleted. However, we found a significant rate of underreporting in the other Tuscan areas (the number of notifications was much lower than the number of hospitalizations). Therefore, we did not considered these data in the analysis but included only the hospitalization data (whose reporting, though not perfect or complete, is more homogenous) in the evaluation. Therefore, even if our results are somewhat limited, they can be considered more accurate and conservative.

The length of stay was not significantly different in the 2 groups. The hospitalizations presumably had the same severity, even though the numbers are different.

Dedicating a specific BCG immunization program to the foreign communities has shown a great degree of efficacy in reducing the clinical and economic impact of TB disease. This effect is particularly evident in children younger than 15 y who have been vaccinated after 2000. This age group showed a lower hospitalization rate than foreign children of the same age living elsewhere in Tuscany, with a ratio about of 1:2. Also, the data regarding the most represented age group indicated a reduction in the number of hospitalizations among the younger children, probably due to vaccination. In future, if this vaccination program is continued, these cohorts will grow up and reduce the prevalence of TB among adults as well, and the notification rate will probably decrease in the LHU4 in Prato, leading to a significant saving of direct and indirect costs. Indirect costs have not been computed in the present study, but were quantified as time loss of 81 days, mainly due to hospitalization (19 days) and additional work days lost (60 days), and amount to an estimated cost of EUR 2,603.²²

Another study ²³ set in Nigeria found that TB-related hospitalizations involved on average 924.98 hours of diagnosis and treatment. Non-hospitalized cases on the other hand needed an average of 141.29 hours. The estimated indirect costs were US$517.98 and US$79.13 for hospitalized and non-hospitalized patient groups, respectively.

Limitations

It was assumed that all vaccines in LHU4 were administered to foreign newborns, since any use by regular newborns would change the numbers and calculation of benefit of the vaccination program.

Having excluded hospitalizations of Italian subjects, the effects could seem over-estimated. An extension of the analysis to Italian subjects would not have shown the same results. Not much can be said about a hypothetical universal BCG vaccination coverage, but this intervention showed a significant gain in health and savings for the foreign population. Therefore, there is indirectly a benefit for Italian residents, because disease, illness and complications are avoided in the foreign beneficiaries.

In the reports of the National Institute of Statistic, only the foreigners who have been regularly registered were considered. However, some TB patients who were not registered were treated under a specific code ‘STP’ (Straniero temporaneamente presente: Foreigner temporally present). This code is given by Italian Sanitary Service Department only in order to be able to deliver sanitary treatments. These irregular foreigners are presumably less numerous than the regular ones, but cannot be counted from the registry offices and consequently not included at denominator. This is one of the limitations of our study because of the some treated patients who have been considered in the numerator, are supposedly STP. Anyway, it is a non-differential bias though STP foreigners are estimated to be present as a small percentage (10%)²⁴ in all groups, it does not alter the results, considering the other conservative aspect of our study.

The use of an anonymous database did nothing to avoid multiple counting of the hospitalization of a single subject. A same case could generate more hospitalisations events and be counted more times. This could lead to an overestimation of the incidence rate. However, these cases of multiple counting are probably very few and can be considered not to have influenced the results.

Costs of domiciliary therapies had to be approximated because a database was not available. We tried to keep our estimate on the low side by considering the best scenario of absence of drug resistances and the possible use of combined and cheaper drugs. Real prices were probably higher, and would have advocated vaccine implementation even more strongly.

Conclusions

In conclusion, the introduction of the BCG immunization program for foreign children in the LHU4 in Prato has had a positive effect in reducing TB impact. It is desirable for this program to be offered in the other LHUs as well, as recommended by the guidelines of the Health Minister.

Methods

Data source

In order to assess the impact of BCG vaccination on hospitalization rate in the LHU4 in Prato, we consulted the regional archive containing records of all discharges from accredited public hospitals in Tuscany in the period 2007–2014, including day-hospitals services, costs, and length of stay. These data are collected and released every year by Regional Statistic Office for statistical or scientific purposes, while withholding the patients' identities.

TB-related hospitalizations were identified from among the discharge records by searching with the following codes in the International Classification of Diseases, 9^th Revision, Clinical Modification (ICD-9-CM, revision 2011) in primary or secondary diagnosis: 010 primary tuberculous infection (0100–0109), 011 tuberculosis of lung infiltrative (0110–0119), 012 tuberculous pleurisy (0120–0128), 013 tuberculous meningitis (0130–0139), 014 tuberculous peritonitis (0140–0149), 015 tuberculosis of vertebral column (0150–0159), 016 tuberculosis of kidney (0160–0169), 017 tuberculosis of skin and subcutaneous cellular tissue (0170–0178), and 018 acute miliary tuberculosis (0180–0189).

Discharge data were analyzed in terms of year of hospitalization, age group, period of stay, and hospitalization costs. Each discharge corresponded to a single hospitalization event and, consequently, because the subjects are anonymous, it is possible that some of them have had multiple hospitalizations.

Only the hospitalization records of foreign patients were selected from the total collected discharge records in Tuscany. Hospitalization of Italian subjects was not considered in the analysis because the BCG vaccination program has not been implemented for this target group. Even though the BCG vaccination campaign was first implemented in 2000 in the LHU4 in Prato, the TB-related discharge records were collected only for the period 2007–2014 because previous data were not easily available and the 8-year period was long enough for the evaluation.

Study population

We have considered 2 different populations for the purpose of comparison of hospitalization rates and related costs: foreigners living in the LHU4 in Prato (and registered at the Local Registry Office) who were vaccinated under the BCG immunization program after 2000, and foreigners living in the other LHUs of Tuscany, registered in the same way and never vaccinated (due to the lack of a specific vaccination program).

Considering the age distribution²⁵ in the population of foreigners living in purview of the LHU4 in Prato and that in the purview of other LHUs during the period of 2007–2014, in Prato the median age was about 30 years, while in the other areas the median age was between 32 and 33 y. The population of foreigners in the purview of the LHU4 in Prato was younger than the population under the other LHUs. The percentage of people aged less than 15 y in Prato was 21.6%, while it was 15.8% in the other LHUs. In both groups, age 0 was the most common age in the group (758/32,245 and 4,507/290,633 respectively). Among the people not younger than 15 years, the most common age was 31 (839/32,245) in the LHU4 in Prato (amounting to 78.4%), and 32 (7,515/290,633) in other LHUs (84.2%). We then disaggregated these populations into 2 age groups: less than 15 y of age and at least 15 y of age. The threshold age of 15 y was chosen because foreign children younger than 15 living in Prato are supposedly vaccinated with the BCG vaccine, while subjects who are 15 y or older were vaccinated neither in Prato nor in the other regional LHUs. Therefore, we could compare subjects who have never been vaccinated with others who have been vaccinated after 2000 (children younger than 15 years). Meanwhile, we could also compare cohorts born before 2000 (15 y or older) that were in the same situation of susceptibility in all of Tuscany because these people were born before the introduction of the BCG vaccine program in the LHU4 in Prato. To correct the possible bias due to different compositions of the 2 populations, we used the direct standardization method, considering the number of foreigners registered at the Registry Offices of LHU4 in Prato and of the other LHUs, disaggregated by age groups. Data regarding the number of foreign inhabitants in Prato and in the other Tuscan areas were collected from National Institute of Statistics (ISTAT) website.²⁵

Design of study

We calculated the average hospitalization rates for each year and for the whole period with these data (number of hospitalizations and population for each year). We applied the hospitalization rate of the 2 populations to the standard population, i.e. the total population of foreigners in Tuscany (sum of the 2 considered populations, disaggregated by age groups). This enabled us to calculate the expected hospitalization rate, if these 2 populations had the same number and age composition of the standard population and to compare the rates. The main outcomes, calculated for the comparison of TB-related hospitalizations among the younger and the older foreigners living in the purview of the LHU4 in Prato and the other LHUs, were the number of expected hospitalizations for each geographic and age group and the Comparative Hospitalization Ratio (CHR). This last ratio is used to compare the standardized hospitalization rates of the 2 populations: if it is approximately one, the 2 populations could be considered similar; if it is more or less than one, this means that the first population at numerator has a lower or greater value of the considered rate respectively.

Costs

The cost analysis was conducted by considering the total costs and average cost per hospitalization for the 2 age groups in the LHU4 in Prato compared to the other LHUs. We calculated the average hospitalization costs for the whole period disaggregated by year and age group. Data regarding costs were available from the regional hospitalisations' database. Costs for domiciliary therapy for each TB-related hospitalization were also added to the average hospitalization costs for the whole period, in order to determine the complete economic impact of TB from the healthcare service perspective. In particular, the domiciliary therapy costs were calculated by considering the average dosage for male or female patients in each age group. Considering that the dosage of therapy depends on the weight of subjects, we assumed an average weight of 70 kg for patients not younger than 15, and an average weight of 25 kg for children younger than 15. This last result was obtained from the growth curve of children.²⁶ The weight in kilograms was considered separately for boys and girls for each year of age; then the average weight for each gender and for the total population was calculated. Considering the average weight, the number of drug packs required for each TB case, and the related costs were calculated. TB domiciliary therapy usually goes on for 6 months and includes isoniazid, rifampin, ethambutol, and pyrazinamide. Combinations and dosages are different during the first 2 months (eradication period) and the last 4 months (maintenance period). The basic therapeutic schedule was calculated in the best scenario, i.e., the absence of drug resistance; in this last case, moxifloxacin, levofloxacin, or streptomycin would be added to the basic treatment instead. We considered the combined formulation for children as well, even though some of the combined drugs (Rimstar®, containing the 4 drugs above) cannot be used for children younger than eight. We still chose this treatment because it was the cheapest, would have not led to an overestimation of the costs of therapy. The length of treatment and dosages were assumed to follow the current protocol of the Infectivology Department in the Santo Stefano Hospital in Prato and to other adopted Italian guidelines.²⁷ Table 6 shows the data of domiciliary therapy and related costs assumed in the economic evaluation of TB impact. Total costs were then calculed considering kind and number of each packages for the whole period and drugs' fee schedule.

Table 6. Assumption on domiciliary therapy and related costs (EUR).

Age group	Domiciliary therapy
Adults (average weight: 70 kg)	For the first 2 months Rimstar ® (combined drug containing isoniazid, rifampin, ethambutol and pyrazinamide), 5 tablets per day, 300 tablets for the whole period, 145.5 EUR in total. Then Rifinah ® (combined drug containing isoniazid and rifampin) for 4 months, 2 tablets per day, 240 tablets for the whole period, 94.8 EUR.Total costs: 240.3 EUR
Children ( average weight: 25 kg)	For the first 2 months Rimstar ®, 2 tablets per day for 2 months, 58.20 EUR. Rifadin syrup ® (rifampin) for 4 months, 12 ml per day, 67.2 EUR in total. Nicozid ® (isoniazid), 6 tablets 200 mg per week for 4 months, 21.39 EUR Total costs: 146.39 EUR

Other source of data

We also asked for the notifications of TB cases reported to the Statistical Regional Office. These TB cases also include cases where the patient was treated at home and not hospitalized. These data were used to evaluate the underreporting rate.

Lastly, in order to calculate the vaccination coverage, we considered the total number of BCG vaccine doses administered in the LHU4 in Prato, sourced from the registers of the Public Health Department, and the number of foreign infants younger than one year living in Prato.¹⁶ However, the coverage calculated in this way is supposedly underestimated, because some foreigners are not eligible for BCG immunization.

Microsoft Office Excel 2007 was used for data analysis.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

We wish to express our sincere thanks to Silvia Callaioli and Lucia Pecori, who provided us all the necessary data from Regional archive.

We are also grateful to the whole staff of the Public Health Department of Prato for their help and support (Lugi Ricci, Susanna Ricci, Luana Paliaga, Cristiana Berti, Maria Pia Nuti, Manuela Gemmi, Vanna Margheri, Sara Pugi, Annalisa Roselli and Carolina Degl'Innocenti).

We take this opportunity to express gratitude also to Donatella Aquilini, head of the Infectivology Department of the S. Stefano Hospital in Prato, for her prompt advice regarding treatment and management of TBC patients and the doctor Lin I Liang for her guidance and teaching during BCG administration.