The European Journal of Public Health Advance Access originally published online on July 13, 2005
The European Journal of Public Health 2005 15(4):343-349; doi:10.1093/eurpub/cki010
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Infectious Diseases |
Determinants of health system delay among confirmed tuberculosis cases in Spain
M. Díez1, M.J. Bleda1, J. Alcaide2, C. Castells3, J.I. Cardenal4, A. Domínguez2, P. Gayoso5, G. Guitiérrez6, C. Huerta7, M.J. López8, T. Moreno9, F. Muñoz10, A. García-Fulgueiras11, M. Picó12, F. Pozo13, J.R. Quirós14, F. Robles15, J.M. Sánchez16, H. Vanaclocha17, T. Vega18 for the Multicentre Project for Tuberculosis Research (MPTR) Study Group
1 TB Research Unit, National Centre for Epidemiology, Carlos III Institute of Public Health, Madrid, Spain
2 Directorate-General of Public Health, Catalonian Regional Authority, Barcelona, Spain
3 Regional Health Authority, Basque Country Regional Authority, Bilbao, Spain
4 Directorate-General of Public Health, Extremaduran Regional Authority, Badajoz, Spain
5 Cristal-Piñor Hospital Complex, Orense, Spain
6 Directorate-General of Public Health, Castile-La Mancha Regional Authority, Toledo, Spain
7 Spanish Pharmaco-epidemiological Research Centre, Madrid, Spain
8 Directorate-General of Public Health, La Rioja Regional Authority. Logroño, Spain
9 Nursing Research Co-ordination & Development Unit, Carlos III Institute of Public Health. Madrid, Spain
10 Valme University Teaching Hospital, Seville, Spain
11 Epidemiology Department, Regional Authority for Health & Consumer Affairs, Murcia, Spain
12 Directorate-General of Public Health, Andalusian Regional Authority, Seville, Spain
13 Doce de Octubre Hospital, Madrid, Spain
14 Directorate-General of Public Health, Principality of Asturias, Oviedo, Spain
15 Regional Health Authority, Melilla Regional Authority, Melilla, Spain
16 Regional Authority for Health, Consumer Affairs & Sports, Ceuta Regional Authority, Ceuta, Spain
17 Directorate-General of Public Health, Valencian Regional Authority, Valencia, Spain
18 Directorate-General of Public Health & Health Care, Castile & León Regional Authority, Valladolid, Spain
Correspondence: Dr Mercedes Díez, Unidad de Investigación en TB, Instituto de Salud Carlos III, C/Sinesio Delgado 6, 28029 Madrid, Spain, tel. +34 91 3877517, fax +34 91 3877872, Email: mdiez{at}isciii.es
Received July 3, 2003, accepted April 6, 2004
 |
Abstract |
|---|
 Top Abstract Materials and methodsResultsDiscussionReferences |
|---|
Background: Health system delay (HSD) is an important issue in tuberculosis (TB) control. This report investigates HSD and associated factors in a cohort of Spanish culture-confirmed TB patients. Methods: Data were collected from clinical records. Using logistic regression with two different cut-off points to define HSD (median and 75th percentile), adjusted odds ratios were used to estimate the association between HSD and different variables. Results: A total of 5184 culture-confirmed TB cases were included. Median and 75th percentile HSD were 6 and 25 days respectively. HSD significantly greater than the median was associated with: age >44 years, past or present intravenous drug use, diagnosis at a primary-care centre, prior preventive therapy, positive histology, request for drug-sensitivity testing, presence of silicosis or neoplasia in addition to TB, presence of non-TB related symptoms, and gastrointestinal site. HSD greater than the 75th percentile was related to the same variables, with the exception of diagnosis at a primary-care centre, positive histology, silicosis, non-TB-related symptoms and gastrointestinal site, for which the association disappeared; in contrast, an association with female gender emerged. Conclusion: Despite free health care being universally available in Spain, there are some groups of TB patients whose treatment is unduly delayed.
Keywords: female TB cases, health system delay, HIV-related TB, intravenous drug user (IDU)
Diagnostic delay in tuberculosis (TB) has been defined as the duration of time from date of symptom onset to institution of treatment.1 It has two constituent components: (i) patient delay, namely, the time that elapses between a patient's feeling ill and first seeking medical attention and (ii) health system delay (HSD), i.e., the time lag between the first medical visit and initiation of anti-TB treatment. Diagnostic delay is an important issue in TB control. On the one hand, delay may be an important factor in the patient's prognosis,2 on the other, long delays mean increased likelihood of TB transmission, both in the community and within health-care facilities. Indeed, HSD has been reported to be an important factor in the occurrence of nosocomial outbreaks of TB.3,4
Spain ranks second in terms of TB incidence5 among European Union countries, and outbreaks of nosocomially transmitted TB have been reported6,7 in which HSD, along with obvious breaks in the infection control measures, were shown to play an important role. However, information on HSD and its determinants is scarce and limited to small samples.8–12
In the course of the Multicentre Project for Tuberculosis Research (MPTR)—a study aimed at describing the epidemiological pattern of TB in Spain as well as TB-related clinical practice13— diagnostic delay was analysed. The median and 75th percentile times in days between a patient's feeling ill and first seeking medical attention were 22 and 57 days, respectively, and a thorough analysis of the factors associated with patient delay greater than either of these parameters has already been published.14 This paper focuses on HSD and its determinants. Since the factors influencing HSD may vary according to whether cases are culture-confirmed TB or not, for this article only results from culture-confirmed cases were considered.
|
Materials and methods |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionReferences |
|---|
TB cases diagnosed during the period May 1996 to April 1997 in 13 (out of the 17) of Spain's Autonomous Regions were identified by active monthly search of seven databases, and detailed clinical and epidemiological information was subsequently collected from their clinical records following standard procedures. The study covered almost 67% of the total Spanish population. A case was defined as anyone who had: (i) either a culture positive for or a direct microscopy compatible with M. tuberculosis and (ii) had been prescribed at least two anti-TB drugs (cases without any positive bacteriological test were only included if the prescription lasted at least 3 months). Persons with a positive culture who had died before receiving treatment were also deemed to be cases. Descriptive results and full details of the MPTR methodology have been published elsewhere.13,15
Dates of first medical visit and initiation of treatment were obtained from the clinical record in the visit that lead to diagnosis and the time lag between these two dates was calculated in days. Only the record kept in the medical facility where diagnosis was made was reviewed, as it was not possible to review several records given the sample size and the broad geographical distribution of the patients. Diagnosis was made in hospitals in 4249 (82%) of cases and in other types of facilities in the remaining cases. Clinical records were reviewed in a standardized way by physicians who followed a training session for this purpose. In addition, strict quality control procedures were implemented to increase data reliability. A manual with precise variable definitions and data collection procedures was developed and double data collection performed monthly in a random sample by the person responsible for the study in each centre. Patients' first contact with the health care system (whether primary care or other) to seek medical attention because of their TB was deemed to be the first medical visit; for patients under frequent medical review, such as those HIV-infected, the visit where the patient showed TB symptoms was considered to be the first visit. The first date that treatment was prescribed was considered to be the date of treatment initiation. All cases with inconsistent or missing information as to date of first medical visit, plus patients who had died before treatment onset (108), were excluded from the analysis. The distribution of HSD, as well as its median and 75th percentile in days, was calculated for patients overall and by subgroup. Association of HSD with different variables previously shown to be related to it was estimated using odds ratios (OR) with their 95% confidence intervals (CI). Multivariate analysis was performed using logistic regression with two different cut-off points (median and 75th percentile) to define HSD. All variables proving significant at a univariate level, as well as any interactions likely to be present, were included in the initial logistic regression models. To decide which variables were to remain in the final models, we used a backward elimination procedure and followed the model-building strategies recommended by Hosmer and Lemeshow.16 All multivariate analysis results were adjusted for regional variations. All analyses were performed using the STATA computer software programme (version 6.0).17
|
Results |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionReferences |
|---|
Of the total of 10 053 TB cases identified in the MPTR,13 108 died before starting treatment and 3730 were not culture positive, thus leaving 6215 potentially eligible study subjects; of these, 1030 had incomplete data [with respect to date of first medical visit (997) or other variables (34)] and one additional case showed inconsistencies, and for these reasons were excluded from the analysis. The remaining 5184 subjects were included in the analysis. No significant differences in demographic variables were found between the 6215 eligible subjects and the final study population except in the percentage of asymptomatic patients (7.5% versus 6.2%, P=0.03) (table 1).
|
Median and 75th percentile delay between date of first medical visit and date of treatment initiation were 6 and 25 days, respectively (5th and 95th percentiles being 0 and 94 days). HSD broke down as follows: 55.6% of the patients initiated treatment within the first week after seeking medical attention, 12.9% in the second week, 8.9% in the third/fourth week, and in 22.7% of cases more than one month passed before treatment was initiated.
Univariate logistic regression models revealed statistically significant differences in HSD according to several variables, both when using the median and the 75th percentile as cut-off points to define HSD (table 2).
|
Results from the multivariate analysis showed that factors significantly associated with HSD greater than the median were: age >44 years, diagnosis at a primary-care centre, intravenous drug use, history of prior chemo-prophylaxis, presence of silicosis or neoplasia in addition to TB, existence of non-TB-related symptoms, request for drug-sensitivity testing (DST), positive results in the histological tests, and gastrointestinal site (table 3).
|
Factors significantly associated with HSD less than the median were: homeless, foreign-national or health-worker status; hospital admission; history of contact with a TB case; presentation with respiratory (versus systemic) TB symptoms; positive sputum smear, abnormal chest X-ray or positive adenosindeaminase (ADA) test; and presence of TB at lymphatic or central nervous sites.
When the 75th percentile was used as the cut-off point to define HSD in the logistic regression model, the factors associated with greater delay were similar to those found when the median was used as the cut-off, though female gender emerged as a factor related to delay greater than the 75th percentile, and other variables lost their statistical significance, possibly owing to a lack of statistical power (table 4). Insofar as factors associated with smaller delay were concerned, the results were again very similar. However pleural and disseminated sites and HIV infection also emerged as variables associated with a shorter delay.
|
|
Discussion |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionReferences |
|---|
Our study showed median HSD to be the same as that found in 1998 in surveillance data in the Autonomous Region of Valencia11 but less than the 15 days reported for Asturias10 and far less than the delay observed in a 1994 study performed at a Valencia city hospital, where only 29% of HIV-negative TB cases initiated treatment less than one month after the first visit, although the corresponding figure among HIV-positive cases was 95%.9 A possible explanation for the differences could be that those studies were performed much earlier than ours, at a time when awareness about TB in Spain was less marked. Furthermore, a degree of recall or interviewer bias was possibly present in the Valencia study, since both cases and interviewers were aware of the diagnosis when data were obtained; moreover, the study included not only culture-confirmed cases, but also those whose diagnosis was based on positive histological or ADA tests.
In comparison with studies conducted outside Spain, HSD proved quite similar to that found in Turkey,18 and less than that encountered elsewhere.19–25 This is not surprising in the case of countries with scarce resources but is unexpected in the case of places such as New York or Australia, where median HSD was 15 and 22 days, respectively.24,25 Several factors may account for these differences: firstly, TB incidence is higher in Spain and it is therefore likely that there is a correspondingly higher index of suspicion among physicians; secondly, Spanish patients are younger, and that makes the diagnosis of TB less complicated since co-morbidity is less frequent in young people; thirdly, the proportion of HIV-related TB cases in our study was almost 18%, and these patients are subjected to close medical follow-up and to a high index of suspicion since, here in Spain, TB is the most common AIDS-defining disease (ADD), with 36.1% of all AIDS cases reported in 1996 having TB as their first ADD;26 finally, the Spanish health system allows for as many medical visits as requested (whether by the patient or the physician) and as many tests as the practitioner deems necessary, at no cost to the patient.
Our results appear to indicate that HSD is the result of a complex array of variables, which could be classified under three headings, namely (i) socio-demographic variables, (ii) variables relating to the provision of heath care and (iii) clinical variables.
The finding of greater HSD linked to rising age and female gender has also been reported in other countries.22,25,27 In the context of an industrialized country such as Spain, these results are probably explained, in the case of older people, by the greater frequency of other medical conditions (which render differential diagnosis more difficult), and, in the case of women, by a lower index of suspicion than that for men.
With respect to homelessness and foreign-nationality, our results differ from those of other studies,24 in the respect that we observed a shorter HSD for patients with than for those without these characteristics. In our view these findings are the result of two factors: firstly, both homeless persons and immigrant groups tend to present with more advanced disease (63.2% of homeless versus 50.3% of non-homeless subjects had a positive sputum smear, with similar results being obtained for non-Spanish versus Spanish nationals); secondly, among medical practitioners the index of suspicion vis-à-vis such patients is higher. This latter reason, together with improved access to medical care, could also account for the lower HSD among health-care personnel.
Most results involving clinical variables (such as the shorter HSD for patients presenting with respiratory symptoms or those with positive results in the diagnostic tests) are what could be expected as a result of the diagnostic process and are common to other studies.24,25
In contrast, what was far more surprising was to find a longer HSD among patients with a positive histological test, a request for DST, a history of prior chemoprophylaxis and among those who were intravenous drug users (IDUs). The first finding could be due to the fact that a positive result in a histology test is an indication of disease at an infrequent extrapulmonary site for which the index of suspicion is low. With regard to DST, it is not standard practice in Spain for DST to be performed on all patients (only 38.5% of those in our study were tested): hence, DST is not available in many facilities where TB is diagnosed and, consequently, specimens have to be sent to other centres, something that inevitably delays the results and could explain the longer HSD.28 The delayed initiation of anti-TB treatment in patients with a history of chemoprophylaxis was also observed in New York24 and might be a reflection of overconfidence on physicians’ part as to the extent to which people complete preventive therapy. Lastly, the longer HSD seen in our data for IDUs could be due to differential treatment of this population, as they are a well-known risk group for TB in Spain29 and entitled to free medical care.
The main limitation of our study is that it was not purpose-designed to study HSD. Hence, patients were questioned on the subject in a way that was no different to the standard pattern adopted in routine clinical practice, and data were collected from cases’ clinical records together with the rest of the information required. It is thus possible that HSD might be underestimated in a case where the doctor taking the clinical history or the person collecting the study data wrongly recorded the date of first consultation as that on which the diagnosis was made rather than the true date of first consultation. To avoid such bias, researchers were specifically trained to collect data in a valid manner, and strict quality-control procedures were implemented throughout the study.13,15 Where TB diagnosis was not made at the first medical visit, another possible limitation could be that the results from diagnostic tests were due to the HSD; in this case, the test results would be indicators rather than predictors of HSD. Indeed, this is something that might affect all retrospective studies, as it is not easy for symptoms present at a first medical visit to be distinguished retrospectively from those noticed when the definitive TB diagnosis is made. To asses to what extent this problem might have influenced our results, we repeated the whole analysis but this time excluded the test-result variables: all factors associated with delay greater than the median or the 75th percentile remained in the models and the magnitude of their effect did not change to any significant degree (results available upon request).
In the course of a multicentre study designed to investigate TB diagnosis and treatment in Spain, detailed clinical and epidemiological information was collected on a large number of culture-confirmed TB cases, thus allowing for a thorough analysis of HSD and its determinants. To better understand HSD, separate analyses were performed using the median and 75th percentile of its distribution as cut-off points to define HSD. In a country such as Spain, i.e. one with free medical care and good availability of diagnostic resources, HSD should be determined by disease presentation and results of diagnostic procedures. Our results suggests that, while this is largely the case, variables such as age, gender, use of illegal drugs and history of preventive therapy seem to play a role.
Key points
|
|
Acknowledgments |
|---|
This work was funded by a grant (99/0016) from Spain's Health Research Fund, Fondo de Investigación Sanitaria (FIS). Dolores Guerra Pérez was awarded two grants by the Carlos III Institute of Public Health (96/4186 and 97/4040). Members of the Multicentre Project for Tuberculosis Research (MPTR) Study Group are listed in: National Centre for Epidemiology Server/TB Research Unit at http://193.146.50.130/uit/Grupos de Trabajo (Working Groups).htm
|
References |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionReferences |
|---|
1 Mori T, Shimao T, Sin BW, Kim SJ. Analysis of case-finding process of tuberculosis in Korea. Tuberc Lung Dis 1992;73:225–31.[Medline]
2 Zafran N, Heldal E, Pavlovic S, et al. Why do our patients die of active tuberculosis in the era of effective therapy? Tuberc Lung Dis 1994;75:329–33.[CrossRef][ISI][Medline]
3 Nosocomial Transmission of Multidrug-resistant tuberculosis among HIV-infected persons-Florida and New York, 1988–1991. MMWR 1991;40:585–91.[Medline]
4 Greenaway C, Menzies D, Fanning A, et al and the Canadian Collaborative Group in Nosocomical Transmission of Tuberculosis. Delay in diagnosis among hospitalized patients with active tuberculosis-Predictors and outcomes. Am J Respir Crit Care Med 2002;165:927–33.
5 EuroTB (InVS/KNCV) and the national coordinators for tuberculosis surveillance in the WHO European Region. Surveillance of Tuberculosis in Europe. Report on tuberculosis cases notified in 1999, March 2002.
6 Multidrug. Multidrug-resistant tuberculosis outbreak in an HIV ward in Madrid, Spain: 1991–1995. MMWR 1996;45:330–3.[Medline]
7 Guerrero A, Cobo J, Fortún J, et al. Nosocomial transmission of Mycobacterium bovis resistant to 11 drugs in people with advanced HIV-1 infection. Lancet 1997;350:1738–42.[CrossRef][ISI][Medline]
8 Morales MM, Suárez-Varela A, Llopis González SA, et al. Delay in childhood tuberculosis detection as a negative factor in the fight against TB. Rev Clin Esp 1992;191:463–7.[Medline]
9 Franco J, Blanquer R, Flores J, et al. Analysis of diagnostic delay in tuberculosis. Med Clin (Barc) 1996;107:453–7.[Medline]
10 Molejón Iglesias A. Diagnostic delay in respiratory tuberculosis in Asturias Health District I (1991–1996). Revista de Formación e Investigación del Hospital de Jarrio 1997;1:15–6.
11 Tuberculosis in the Valencian Region: 1998. Informes de Salud, No. 61. Conselleria de Sanitat. Valencia: Generalitat Valenciana, 2001.
12 Tuberculosis Report 2000. Centro de Prevención y Control de la Tuberculosis de Jerez. Jerez: Servicio Andaluz de Salud, 2001.
13 Diez M, Huerta C, Moreno T, et al for the MPTR Study Group. Tuberculosis in Spain: epidemiological pattern and clinical practice. Int J Tuberc Lung Dis 2002;6:295–300.[Medline]
14 Díez M, Bleda MJ, Alcaide J, et al. (Proyecto Multicéntrico de Investigación sobre Tuberculosis- PMIT). Determinants of patient delay among tuberculosis cases in Spain: Eur J Public Health 2004;14:151–5.
15 MPTR Working Group. Tuberculosis in Spain: Results of the Multicentre Project for Tuberculosis Research—MPTR. Madrid: Instituto de Salud Carlos III, 2000.
16 Hosmer DW, Lemeshow S. Applied Logistic Regression. Chapter 4. Wiley-Interscience, 1989.
17 StataCorp. Stata Statistical Software: Release 6.0. College Station, TX: Stata Corporation, 1999.
18 Yilmaz A, Boga S, Sulu E, et al. Delays in the diagnosis and treatment of hospitalized patients with smear-positive pulmonary tuberculosis. Respir Med 2001;95:802–5.[Medline]
19 Kramer F, Modilevsky T, Waliany AR, et al. Delayed diagnosis of tuberculosis in patients with human immunodeficiency virus infection. Am J Med 1990;89:451–6.[CrossRef][ISI][Medline]
20 Beyers N, Gie RP, Schaaf HS, et al. Delay in diagnosis, notification and initiation of treatment and compliance in children with tuberculosis. Tuberc Lung Dis 1994;75:260–5.[CrossRef][ISI][Medline]
21 Steen TW, Mazonde GN. Pulmonary tuberculosis in Kweneng District. Botswana: delays in diagnosis in 212 smear-positive patients. Int J Tuberc Lung Dis 1998;2:627–34.
22 Lawn SD, Afful B, Acheampong JW. Pulmonary tuberculosis: diagnostic delay in Ghanaian adults. Int J Tuberc Lung Dis 1998;2:635–40.[Medline]
23 Lönnroth K, Thuong LM, Linh PD, Diwan L. Delay and discontinuity- a survey of TB patients’ search of a diagnosis in a diversified health care system. Int J Tuberc Lung Dis 1999;3:992–1000.[ISI][Medline]
24 Sherman LF, Fujiwara PI, Cook SV, et al. Patient and health care system delays in the diagnosis and treatment of tuberculosis. Int J Tuberc Lung Dis 1999;3:1088–95.[ISI][Medline]
25 Ward J, Siskind V, Konstantinos A. Patient and health care system delays in Queensland tuberculosis patients, 1985-1998. Int J Tuberc Lung Dis 2001;5:1021–7.[Medline]
26 Castilla Catalán J, Guerra Romero L, Cañón Campos J, et al. Decrease in tuberculosis incidence following the introduction of new HIV treatments. Rev Clin Esp 1999;199:76–7.
27 Needham DM, Foster SD, Tomlinson G, Godfrey-Fausset P. Socio-economic, gender and health services factors affecting diagnostic delay for tuberculosis patients in urban Zambia. Trop Med Int Health 2001;6:256–9.[ISI][Medline]
28 Krishnan S, Koehler JE, Benjamin R, Reingold A. Reducing delays in the laboratory diagnosis of tuberculosis (letter). Am J Public Health 1996;86:1171.[Medline]
29 Cobo J, Oliva J, Asensio A, et al. Predicting tuberculosis among HIV-infected patients admitted to hospital:comparison of a model with clinical judgement of infectious disease specialists. Eur J Clin Microbiol Infect Dis 2001;20:779–84.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||