Results: Defaults were 25% and 45% in CAT I & CAT II respectively. ... (AOR=2.55(1.31-4.94)-Cat II) were predictors of default among re-treatment patients.
Original Article
Ind. J Tub., 2003, 50,185
DEFAULTS AMONG TUBERCULOSIS PATIENTS TREATED UNDER DOTS IN BANGALORE CITY : A SEARCH FOR SOLUTION* Sophia Vijay1 VH Balasangameswara2, P S Jagannatha3, VN Saroja4 and P Kumar5 Summary: Setting: Revised National Tuberculosis Control Programme emphasizing on DOTS in the urban set-up of Bangalore Mahanagar Palike. Objective: 1) To identify socio-demographic and treatment related risk factors predictive of default with DOT 2) To study treatment regularity and final bacteriological profile of defaulted tuberculosis (TB) patients. Design: A retrospective case control analysis of the data obtained from the cohort study of 264 new (CAT I) and 219 retreatment (CAT II) bacteriological positive patients treated under DOTS from March 99 to September 2000 and followed up till treatment outcome. Data collection was through interviews at treatment initiation and at treatment outcome using pre-tested semi-structured interview schedules. Results: Defaults were 25% and 45% in CAT I & CAT II respectively. The predictive factors associated with default identified through multivariate logistic regression were male (Adjusted Odds Ratio (AOR) =2.49(1.10-6.18)-CAT I, 2.78(1.15-6.7)-Cat II) and alcoholism (AOR=6.38 (3.25-12.5)-CAT I, 3.93 (2.1-7.5)-Cat II). In addition, patients having poor knowledge of TB (AOR=3.06(1.24-7.54)-Cat II) and those returning for treatment after default (AOR=2.55(1.31-4.94)-Cat II) were predictors of default among re-treatment patients. Majority (CAT I =65.7%, CAT II =71%) of the patients defaulted in the Intensive Phase particularly after the 12th dose. More than half of the defaulted patients remained bacteriologically positive at the end of treatment period. Conclusion: The males and alcoholics are predictive risk factors of default with DOT in an urban setting. Those returning for treatment after default and having poor knowledge of disease are additional risk factors among retreatment patients. Devoting attention to those at potential risk of default from the initiation of treatment with close supervision and repeated counselling would be a major input to minimize defaults and achieve desired goal of RNTCP. Key words: Default, DOT, Case-control study
INTRODUCTION Defaulting from treatment has been one of the major obstacles to treatment management and an important challenge for TB control1,2. Inability to complete the prescribed regimen which is quite common in self-administered treatment3, is an important cause for treatment failure, relapse, acquired drug resistance and on-going transmission of infection4. The consequences of default could be disastrous particularly in the context of intermittent Short Course Chemotherapy (SCC) regimens. Over the years there has been increasing emphasis on Directly Observed Treatment short course (DOTS) to ensure treatment adherence, wherein each dose of treatment is given under the observation of a health worker. The adoption of DOTS has given impressive results with higher treatment success being reported from developing5 and industrialized countries6. Yet,
default continues to occur in certain situations and is a matter of concern. An efficient network of health infrastructure with committed treatment organization is most essential for the success of DOTS. The challenges encountered while implementing DOT vary from place to place depending on the geographic terrain, demographic structure & socio-cultural milieu. The Revised National Tuberculosis Control Programme (RNTCP) based on DOTS strategy was implemented in the country in 1993 and is under rapid expansion since 1998. The major thrust of RNTCP is achieving a cure rate of more than 85% 5. Strict adherence to Directly Observed Treatment is likely to minimize defaults and is therefore essential for the desired treatment success. The National Tuberculosis Institute (NTI)
* Presented at the 57th National Conference on Tuberculosis and Chest Diseases - Goa : 26-29, September, 2002 1. Sr. TB Specialist 2. Chief Medical Officer 3. Statistical Assistant 4. Sr. Public Health Nurse 5. Director Correspondence: Director, National Tuberculosis Institute, 8, Bellary Road, Bangalore – 560 003
Indian Journal of Tuberculosis
186
SOPHIA VIJAY ET AL
had undertaken a prospective cohort study among new and retreatment cases in the metropolis of Bangalore, six months after RNTCP implementation. The treatment success of these cohorts was vitiated by high defaulter rates1. Hence, an in-depth analysis of the data on default available from the above study was warranted to be able to propose suitable and effective operational changes within the RNTCP setup.
collected from the treatment cards. Patients were interviewed at their residence using pre-tested semistructured schedule. Repeated attempts were made to contact defaulted patients. The field team collected sputum samples from patients, three at diagnosis and two at time of final interview. These were processed for bacteriological examination (microscopy, culture and drug susceptibility) in the reference laboratory at NTI.
Objectives of this exercise were: to identify socio-demographic & treatment related risk factors predictive of default with DOT in an urban set up and to study treatment regularity with final bacteriological status of defaulted patients.
After excluding the wrongly categorized and culture negatives, 271 new and 226 re-treatment smear and culture positive patients formed the study cohorts. Statistical Methods
MATERIAL AND METHODS Source of data Data for the analysis were taken from the prospective study of new and re-treatment smear positive patients followed up till treatment outcome6. Three hundred and sixty six (366) new and 269 retreatment smear positive patients were initiated on intermittent SCC regimen CAT I & CAT II given under DOT according to RNTCP policy2 from March to December 1999 and March 1999 to September 2000 respectively in all the 104 treatment centres of Bangalore Mahanagara Palika (BMP). The trained NTI field team interviewed these patients soon after treatment initiation to obtain socio demographic details and particulars regarding previous histories of antiTB treatment, if any. Patients who could read and write were considered as literates and those habituated to alcohol were labeled as Alcoholics. Knowledge of Tuberculosis was assessed through three open-ended questions focused on the cause, infectiousness of disease and its cure. While scoring these responses, maximum weightage was given to the question related to cure. In addition, their awareness on treatment duration, regularity and frequency of drug administration was assessed. Final interviews were conducted on declaration of the treatment outcome to obtain details of current treatment undertaken that included place of DOT, treatment regularity, retrieval actions and reasons for stopping treatment. Relevant information was also
Indian Journal of Tuberculosis
A retrospective case-control analysis was done on the data available from the cohorts to identify risk factors associated with default. They were dichotomized into defaulted (cases) and treatment completed group (controls) after excluding death as treatment out come. Defaulted were those who had stopped treatment for > 2 months consecutively and treatment completed (henceforth referred to as completed) comprised of patients who completed the prescribed treatment regimen with or without interruption. The association between potential sociodemographic and treatment related risk factors among cases and controls was initially studied through univariate analysis. The categorical variables were assessed using χ 2. Odds Ratio (OR) and 95% Confidence Interval (CI) were calculated. To estimate the independent effect of the factors that were significantly associated with default and to control the confounding effect they may have on each other, logistic regression analysis was done. The variables were included if their respective univariate analysis yielded P Rs.633 in comparison to the completed group. The difference in marital status was also statistically significant among new patients, wherein a higher proportion of defaulted patients was among those married in contrast to the completed group. Other predictive variables did not show significant difference between completed and defaulted among Cat-I and Cat-II patients (Table 1).
Socio-demographic factors Treatment related factors In the univariate analysis of sociodemographic characters between the cases and controls, treatment defaults were significantly higher
Among the re-treatment defaulted patients 78.8% belonged to the type ‘Treatment after default’
CAT-I (264) Control (197) (Completed)
CAT-II (219)
Case (67) (Defaulted, 25.4%) Interviewed 40 Migrated 16
Control 120 (Completed)
Not interviewed 27 (40.3%) NA 7
Case (99) (Defaulted, 45.2%) Interviewed 66
Dead 4
Not interviewed 33 (33.3%)
Migrated 17
NA 14
Dead 2
Fig. 1: Study Group
Indian Journal of Tuberculosis
188
SOPHIA VIJAY ET AL
(TAD) which was significantly more than the TAD’s in the completed group (54.2%). Though all patients in the study group were well informed about treatment duration and its regularity, most of them did not have any knowledge regarding the disease. However, the difference between defaulted and completed group was significant only among Cat-II patients wherein >90% of defaulted did not have any knowledge about the disease (Table 2). Very few patients [18 (6.8%) on Cat-I, 16 (7.3%) on Cat-II] received treatment (DOT) by a DOT provider other than the centre staff. Majority approached the centre for DOT. The patients had to travel a distance ranging from 1 Km to 19 Kms to reach the treatment centre. The median distance travelled by the study group was 2 Kms without significant difference between the defaulted and completed group. Retrieval action through home visit was taken for > 80% of the study patients who had missed one or more doses/ collections during the treatment. There was significant difference between the proportions of retrieval action taken between defaulted and completed group among new patients. (Table 2). In the logistic regression analysis, predictive factors that remained independently associated with
Table 1: Socio-Demographic risk factors among treatment completed and defaulted patients CAT I (New, N = 264) Risk factors
Age ≥ median Gender Male Education Literate Employment Employed Per Capita Income < Median Marital status Married Alcoholism Present
CAT II (Retreatment, N = 219)
Compl. N=197
Default N=67
P value
Odds Ratio (95% C.I)
Compl. N=120
Default N=99
P value
Odds Ratio (95% C.I)
88 (44.7)
45 (67.2)
0.001
2.53 (1.42 –4.54)
57 (46.7)
65 (53.3)
0.007
2.11 (1.22-3.66)
126 (64.0)
60 (89.6)
0.00
4.83 (2.1011.13)
82 (68.3)
90 (90.9)
0.000
4.63 (2.1110.17)
141 (71.6)
46 (68.7)
0.650
1.15 (0.63-2.10)
83 (69.2)
66 (66.7)
0.693
1.12 (0.63–1.98)
98 (49.7)
40 (59.7)
0.159
0.67 (0.38-1.17)
66 (55.0)
64 (64.6)
0.148
1.15 (0.87-2.59)
103 (52.3)
34 (50.7)
0.828
0.94 (0.54 –1.64)
42 (42.4)
68 (56.7)
0.036
0.56 (0.33-0.97)
110 (55.8)
49 (73.1)
0.012
2.15 (1.17-3.96)
89 (74.2)
69 (69.7)
0.518
0.80 (0.44-1.44)
20 (13.2)
38 (56.7)
0.00
8.62 (4.5716.27)
33 (32.7)
68 (67.3)
0.000
5.33 (2.98-9.51)
( ) percentage , CI= Confidence Interval
Indian Journal of Tuberculosis
DEFAULTS UNDER DOTS
189
Table 2: Treatment Related risk factors among completed and defaulted patients CAT I (New, N = 264) Risk factors
Knowledge of TB Nil Patient Type TAD # Distance from Centre > 2 Km DOT At Centre Dose Missed Retrieval Action Taken
CAT II (Retreatment, N = 219)
Compl. N=197
Default N=67
P value
Odds Ratio (95% C.I)
Compl. N=120
Default N=99
P value
Odds Ratio (95% C.I)
165 (83.8)
57 (85.1)
0.799
1.11 (0.51-2.39)
95 (79.2)
90 (90.9)
0.017
2.63 (1.17-5.94)
65 (54.2)
78 (78.0)
0.000
3.14 (1.7-5.7)
115 (58.4)
39 (58.2)
0.981
0.99 (0.57-1.74)
73 (60.8)
56 (56.6)
0.523
0.81 (0.47-1.39)
181 (91.9)
65 (97.4)
0.150
2.87 (0.64-12.84)
111 (92.5)
92 (92.9)
0.903
1.07 (0.38-2.97)
103 (52.3)
52 (77.6)
0.0002
0.47 (0.25-0.88)
80 (66.7)
79 (79.8)
0.030
1.98 (1.06-3.67)
97 * (95.1)
56 (83.6)
0.01
5.25 (2.48 – 11.32)
70* (87.5)
89 (89.9)
0.61
6.36 (2.86-14.45)
* Retrieval Action among completed is applicable for those missing doses. # TAD= Treatment After Default Dependent Variable = Outcome (Cases =1 vs Control=0)
Variable
Gender Male/Female Alcoholism Alcoholic / Non Alcoholic Knowledge of TB Nil / Some Patient Type
Table 3: Logistic Regression Model (after backward elimination) New Retreatment Coefficient
S.E
P Value*
0.9138
0.46
0.04
1.8154
0.34
-
-
Coefficient
S.E
P Value*
AOR (95% C.I)
2.49 (1.106.18)
1.02
0.45
0.023
2.78 (1.15 –6.7)
0.00
6.38 (3.2512.5)
1.37
0.33
0.0000
3.9 (2.1-7.5)
-
-
1.12
0.46
0.015
3.1 (1.2-7.5) 2.5
0.94
0.34
0.005
(1.3-4.9)
0.3381
0.36
0.000
TAD /other types Constant % correct predictors
2.5096
0.41
AOR (95% C.I)
0.00 79.17%
70.78%
d.f= degrees of Freedom, AOR= Adjusted Odds Ratio *Note: Wald Statistics are distributed χ2 with 1 d.f
Indian Journal of Tuberculosis
190
SOPHIA VIJAY ET AL
default were gender and alcoholism among new and re-treatment patients. Knowledge of TB and the type of patient (TAD) emerged as additional risk factors associated with default for re-treatment patients. The model predicted 79.1% & 70.8% of observations correctly in new and re-treatment patients. The likelihood test (p=0.9928 & p=0.4290) confirmed that regression model fitted the data well (Table 3). Pattern of Treatment regularity The drug intake of defaulted and completed group was studied through a scatter plot of doses Vs treatment days among the study group (Figs.2 a & b). The expected duration line in the plot is drawn presuming that each dose of IP & CP is taken as per schedule with out any missed doses. Expected & Observed treatment duration in days for new and re-treatment patients is shown in Table 4. More than half the patients in defaulted as well as completed group had missed one or more doses either in IP or CP before ultimately completing or defaulting from
the scheduled treatment. This is reflected in the range of treatment duration and median days within each group (table 4). New cases : Of the 67 total defaults, 44 (65.71%) were in the IP and among these, 31 (70.5%) defaulted after the 12th dose. The pattern of drug intake shown in Fig. 3a suggests treatment irregularity in terms of deviation from the expected line. While 52 (77.6%) of defaulted patients missed one or more doses prior to treatment default, only 102 (51.8%) in the completed group did so during the entire treatment period (Table 2). The difference was highly significant (p
