J. Appl. Environ. Biol. Sci., 5(7)37-42, 2015 © 2015, TextRoad Publication
ISSN: 2090-4274 Journal of Applied Environmental and Biological Sciences www.textroad.com
Meta-Analysis on the Factors Affecting the Occurrence of Pulmonary Tuberculosis Dya Sustarmi1*, Bambang Widjanarko Otok2 1
2
Lecture of College of Health Sciences Hang Tuah, Surabaya, Indonesia Laboratory of Environmental and Health Statistic, Sepuluh Nopember Institute of Technology (ITS), Surabaya, Indonesia Received: March 11, 2015 Accepted: May 20, 2015
ABSTRACT Meta-analysis did not focus on the conclusions derived from various studies, but rather focus on data, such as the magnitude of effect size. Focus on the effect size of the empirical findings of this meta-analysis is an advantage compared to other literature review. By way of synthesizing empirical studies that already exist, meta-analysis can be an efficient tool to gain a thorough conclusion to a phenomenon, without having to conduct new research or falter in aligning theories or different views. The results showed that of the 20 studies in the meta-analysis synthesized binary data there is a relationship between ventilation with pulmonary TB incidence with significant value p=0,000<(α=0,05), but partially 13 significant studies and 7 are not significant. KEYWORDS:Meta Analysis, binary response, pulmonary tuberculosis, effect size 1.
INTRODUCTION
Along with the development of medical research, biostatistics prominent role. Of the various health research conducted separately pursued summarizes the results of similar studies using certain statistical techniques so that the results of such research more informative summary. New trends in research synthesis is quantitatively combine the knowledge obtained from empirical studies on a topic by using meta-analysis. The application of quantitative research synthesis methods are rarely found before the 1970s. The growing popularity of the method after Glass in 1976 introduced the phrase meta-analysis as the statistical analysis of the various analyzes resulting from individual studies with the aim of integrating existing inventions [1]. According to [2], Indonesia as the largest contributor to the TB number 3 in the world after India and China with approximately 539.000 new cases and approximately 101.000 deaths per year. Household Health Survey (Survey) in 1995, placing TB as the third biggest cause of death after cardiovascular diseases and respiratory diseases, and is the largest in the group a number of infectious diseases. [3], Risk factors of pulmonary TB according to Ministry of Health in 2007, among others: environmental factors consist of: ventilation, indoor density, behavioral factors, germ concentration, duration of contact, malnutrition, disease DM, immuno-suppressants, delays in diagnosis and treatment, inadequate management and health conditions. Meta-analysis allows the calculation of the value of the estimated effect size of each study and the effect size is the core analysis. Effect size is a necessary, if a doctor / patient must decide whether they need treatment, they should know whether the treatment reduces the risk of death of 5%, 10% or 20%, and this information can be obtained from the effect size. Narrative review did not have a good mechanism for assessing the consistency of effect size. Meta-analysis allows for combining the results of the diverse and attention to the relative sample size and effect size. Results of this review is accurate given the very broad scope of this analysis and centralized. Meta-analyzes also provide an answer to the contentious issue because of a conflict in the findings of similar studies vary [4][5][6] Meta-analyzes were done properly can provide more definitive information about things reported in the original study, including the effect size is more defined, more narrow confidence intervals, as well as an analysis of the subgroup. One of the advantages of meta-analysis was obtained a new study with a large number of subjects so that it can be concluded that a more definitive. 2. METHODOLOGY Collecting data manually and computerized by searching the collection of primary research results through the internet using the keywords pulmonary tuberculosis [7][8]. Having obtained the results of further research on pulmonary TB sorted again in accordance with the criteria that the research using case control design and 2x2 tables for later quantitative abstraction made of each of the research results [4][5]. Based on the literature on some research on factors affecting the incidence of pulmonary tuberculosis found 20 similar studies that describe environmental factors ventilation is one of the factors that influence the incidence of pulmonary tuberculosis with varying results. Of the 20 studies that made a conclusion through the incorporation of the Corresponding author: Dya Sustarmi,Lecture of College of Health Sciences Hang Tuah,Surabaya, Indonesia Emails:
[email protected], 37
Sustarmi and Otok, 2015
results of research using meta-analysis to calculate effect size to get an answer that ventilation affects the incidence pulmonary tuberculosis [9][10] ..[23]. More detailed steps include: • Determine the topic meta-analysis is the physical home environment factors (ventilation) affect the incidence pulmonary tuberculosis. • Develop research questions: whether the physical environment factors house (ventilation) affect the incidence pulmonary tuberculosis?. • Develop research hypothesis that physical environmental factors house (ventilation) affect the incidence pulmonary tuberculosis. • Determine the purpose and benefits of meta-analysis. • Make research selection criteria to be included in the meta-analysis. • Abstraction quantitative results of each study covers odds ratio, log odds ratios and others. • Combining the results of studies using random effect model and fixed effect models. • Interpret the results • Report the results 3.
of of of of
RESULTS AND DISCUSSION
The study is a meta-analysis of studies that are combining the results of several studies to make a conclusion. With the development of the research, the more studies that are replicas, so more and make the information obtained about the concepts related to other concepts. Documenting the results of the study include the number of cases and controls in each ventilation research on the relationship with the incidence of pulmonary tuberculosis. The results can be seen in the following table: Table 1. Data Results Case Control Ventilation Factor In pulmonary TB incidence NO 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
RESEARCH Kussularso, 1994 Kussularso, 1994 Iman Jaya, 2000 Iman Jaya, 2000 Ahmad Dahlan, 2001 Nur Romaita, 2002 FX. Agus Budiyono, 2003 FX. Agus Budiyono, 2003 Didik Supriyono, 2003 Supriyadi, 2003 Ibupertiwi, 2004 Ike Silviana, 2006 Ike Silviana, 2006 Ike Silviana, 2006 Adrial, 2006 A.H Mahpudin, 2006 Okta Vinia Radhumi, 2007 Helda Suarni, 2009 Asriani, 2009 Nur Yasiroh, 2009
EVENT (+) 94 135 52 29 38 27 70 79 97 21 52 81 72 61 71 14 42 40 24 15
CASE EVENT (-) 108 67 27 50 12 5 18 9 28 2 28 14 23 34 29 62 5 10 7 15
n 202 202 79 79 50 32 88 88 125 23 80 95 95 95 100 76 47 50 31 30
EVENT (+) 106 139 32 24 41 12 63 68 50 129 38 23 18 15 35 40 45 11 5 10
CONTROL EVENT (-) 96 63 47 55 9 20 25 20 75 148 42 72 77 80 65 264 2 39 26 20
n 202 202 79 79 50 32 88 88 125 277 80 95 95 95 100 304 47 50 31 30
Table 1, shows the total number of studies is 20 study, with details Kussularso 2 study, Iman Jaya 2 studies, Ahmad Dahlan 1 study, NurRomaita 1 study, FX. Agus Budiyono 2 studies, Didik Supriyono 1 study, Supriyadi 1 study, Ibupertiwi 1 study, Ike Silviana 3 studies, Adrial 1 study, AH Mahpudin 1 study, Oktavinia Radhumi 1 study, Helda Suarni 1 study, Asriani 1 study and Nur Yasiroh 1 study. Furthermore, the calculation of the effect size of each study and the summary of the entire study to make a conclusion whether the ventilation factors affect the incidence of pulmonary tuberculosis. Analysis of effect size in a meta-analysis of binary data is performed as follows. • Make contingency table and calculate the odds ratio value of each study, for example in study -1: Table 2. Table of Contingency 2 x 2 The incidence of pulmonary tuberculosis TREATED CONTROLLED
Odds _ Ratio =
EVENT (+)
EVENT (-)
A (94) C (106)
B (108) D (96)
AD 94 × 96 = = 0.788 BC 108 × 106
38
J. Appl. Environ. Biol. Sci., 5(7)37-42, 2015
And so on until all 20 studies. Odds ratio results of each study can be seen in the following table: Table 3. Odds ratios of each study STUDY Kussularso.1994 Sularso.1994 Iman.2000 Jaya Iman.2000 Ahmad Dahlan.2001
Odds ratio 0.788 0.913 2.829 1.329 4.557
Lower limit 0.533 0.601 1.482 0.685 2.128
Upper limit 1.165 1.387 5.399 2.579 9.760
Z-value -1.193 -0.426 3.152 0.842 3.903
p-value 0.884 0.665 0.001 0.200 0.000
Nur R.2002
9.000
2.730
29.667
3.610
0.000
Agus.2003
1.543
0.770
3.092
1.224
0.111
Budiono.2003
2.582
1.102
6.046
2.185
0.014
Didik S.2003
5.196
2.992
9.026
5.850
0.000
12.047
2.771
52.368
3.319
0.000
Supriyadi.2003
2.053
1.087
3.875
2.219
0.013
Ike.2006
18.112
8.673
37.823
7.710
0.000
Silviana.2006
13.391
6.680
26.845
7.312
0.000
Ike Silviana.2006
9.569
4.786
19.132
6.389
0.000
Adrial.2006
4.547
2.505
8.253
4.979
0.000
AH Mahpudin.2006
1.490
0.764
2.908
1.170
0.121
Okta V.2007
0.373
0.069
2.029
-1.141
0.873
Helda.2009
14.182
5.412
37.161
5.396
0.000
Asriani.2009
17.829
4.983
63.786
4.429
0.000
Nur Y.2009
2.000
0.705
5.677
1.302
0.096
Ibupertiwi.2004
• calculate the value of the log odds ratio, Variance and Standard Error of each study each study, for example in study -1: 1 1 1 1 logOdds _ Ratio = ln(0.788) = −0.238 , VlogOdds _ Ratio = + + + = 0.040 94 108 106 96
SElogOdds _ Ratio = VlogOdds _ Ratio = 0.040 = 0.199 And so on until all 20 studies. Results of the log odds ratio, Variance and Standard Error of each study can be seen in the following table: Table 4. Log Odds Ratio, Variance and Standard Error of each study STUDY Kussularso.1994 Sularso.1994 Iman.2000 Jaya Iman.2000 Ahmad Dahlan.2001 Nur R.2002 Agus.2003 Budiono.2003 Didik S.2003 Supriyadi.2003 Ibupertiwi.2004 Ike.2006 Silviana.2006 Ike Silviana.2006 Adrial.2006 AH Mahpudin.2006 Okta V.2007 Helda.2009 Asriani.2009 Nur Y.2009
Log Odds Ratio -0.238 -0.091 1.040 0.285 1.517 2.197 0.434 0.948 1.648 2.489 0.719 2.897 2.595 2.258 1.514 0.399 -0.985 2.652 2.881 0.693
Standard error 0.199 0.213 0.330 0.338 0.389 0.609 0.355 0.434 0.282 0.750 0.324 0.376 0.355 0.354 0.304 0.341 0.864 0.491 0.650 0.532
Variance 0.040 0.045 0.109 0.114 0.151 0.370 0.126 0.188 0.079 0.562 0.105 0.141 0.126 0.125 0.093 0.116 0.746 0.242 0.423 0.283
39
Lower limit -0.629 -0.508 0.393 -0.378 0.755 1.004 -0.261 0.098 1.096 1.019 0.084 2.160 1.899 1.566 0.918 -0.270 -2.678 1.689 1.606 -0.350
Upper limit 0.153 0.327 1.686 0.947 2.278 3.390 1.129 1.799 2.200 3.958 1.354 3.633 3.290 2.951 2.111 1.068 0.708 3.615 4.156 1.736
Z-value -1.193 -0.426 3.152 0.842 3.903 3.610 1.224 2.185 5.850 3.319 2.219 7.710 7.312 6.389 4.979 1.170 -1.141 5.396 4.429 1.302
pvalue 0.884 0.665 0.001 0.200 0.000 0.000 0.111 0.014 0.000 0.000 0.013 0.000 0.000 0.000 0.000 0.121 0.873 0.000 0.000 0.096
Sustarmi and Otok, 2015
• heterogeneity test to get a variance of each study is homogeneous or heterogeneous example in study -1: Heterogeneity test is done by calculating T2as follows: 25114.138 = 176.741 = 328.571 − 165.410 df = 20 – 1 = 19 2079.192 = 152.840 = 165.410 − 165.410 176.741 − 19 ଶ = = 1.032 152.840 Having obtained value T2 then calculated summary effect by using random effect model as follows: V = 0.040 + 1.032 = 1.072 1 = = 0.933 1.072 And so on for all 20 studies, the results of the calculation are presented in the following table: Table 5. Random Effect Model of Each Study Study
Y
Kussularso.1994
-0.238
Variance within 0.040
Variance between 1.032
Variance total 1.072
W
WY
0.933
-0.222
Sularso.1994
-0.091
0.045
1.032
1.077
0.928
-0.084
Iman.2000
1.040
0.109
1.032
1.141
0.877
0.911
Jaya Iman.2000
0.285
0.114
1.032
1.146
0.872
0.248
Ahmad Dahlan.2001
1.517
0.151
1.032
1.183
0.845
1.282
Nur R.2002
2.197
0.370
1.032
1.402
0.713
1.567
Agus.2003
0.434
0.126
1.032
1.158
0.864
0.375
Budiono.2003
0.948
0.188
1.032
1.220
0.819
0.777
Didik S.2003
1.648
0.079
1.032
1.111
0.900
1.483
Supriyadi.2003
2.489
0.562
1.032
1.594
0.627
1.561
Ibupertiwi.2004
0.719
0.105
1.032
1.137
0.879
0.632
Ike.2006
2.897
0.141
1.032
1.173
0.852
2.469
Silviana.2006
2.595
0.126
1.032
1.158
0.864
2.241
Ike Silviana.2006
2.258
0.125
1.032
1.157
0.864
1.952
Adrial.2006
1.514
0.093
1.032
1.125
0.889
1.347
AH Mahpudin.2006
0.399
0.116
1.032
1.148
0.871
0.347
Okta V.2007
-0.985
0.746
1.032
1.778
0.562
-0.554
Helda.2009
2.652
0.242
1.032
1.274
0.785
2.082
Asriani.2009
2.881
0.423
1.032
1.455
0.687
1.980
Nur Y.2009
0.693
0.283
1.032
1.315
• calculatesummary effect: o random effect model Based on Table 4 obtained ∑W = 16.394 and ∑WY = 20.922. 20.922 = = 1.276 16.394 1 = = 0.061 16.394 = √0.061 = 0.247 LLm = 1.276 – (1.96 x 0.247) = 0.792
40
0.760
0.527
16.394
20.922
J. Appl. Environ. Biol. Sci., 5(7)37-42, 2015
LLOR = Exp (0.792) = 2.208 ULm = 1.276 + (1.96 x 0.247) = 1.760 ULOR = Exp (1.760) = 5.814 1.276 5.167 0.247 OR = Exp (5.167) = 3.583 p = 1 – (1.000) = 0.000 o
fixed effect model summary Odds Ratio = 3.583 (2.208 – 5.814). z = 5.167 and p = 0.000
Results of analysis of binary data meta-analysis using software CMA.2 presented in graphical form as shown below.
Based on the above image can be seen there are 20 studies were combined in a meta-analysis of binary data with significant value p=0.000<(α=0.05) in this case statistically H0 rejected, meaning that there is a relationship between ventilation with the incidence of pulmonary tuberculosis. 4.
CONCLUSION
Results of the study are combined in a meta-analysis is the number of cases and controls pulmonary TB caused by factors ventilation. OR calculation results of each study were incorporated from 20 studies, there were 13 significant research and 7 research insignificant. OR is a summary calculation result means that there is a significant relationship ventilation with the occurrence of pulmonary TB. REFERENCES [1] King William., (2005) Understanding the Role and Methods of Meta-Analysis in IS Research, Communication of the Association for Information Systems. [2] Depkes RI., (2007) Pedoman Nasional Penanggulangan Tuberculosis, Edisi 2 cetakan pertama, Departemen Kesehatan Republik Indonesia, Jakarta. [3] Dahlan A., (2001) Faktor-Faktor Resiko Lingkungan Yang Berhubungan dengan Kejadian Penyakit TB Paru BTA (+) (Studi Kasus Kontrol) di Kota Jambi tahun 2000-2001, Tesis, FKM-Universitas Indonesia. [4] Anwar R., (2005) Meta Analisis, Fakultas Kedokteran Universitas Padjajaran Bandung
41
Sustarmi and Otok, 2015
[5] Borenstein M., Larry V Hedges., Julian P.T Higgins., Hannah R Rothstein., (2009) Introduction to Meta-Analysis. United Kingdom John Wiley & Sons, Ltd [6] Hamer R.M., Pippa M.S., (2005).SAS@Tools for Meta Analysis, The University of North California, North California. [7] Levy, P.S., and Stanley, L. (1999). Sampling of Populations: Methods and Applications. Third Edition. John Wiley and Sons. Inc. New York. [8] Aswar A., (1995) Pengantar Ilmu Kesehatan Lingkungan, Jakarta: Mutiara Sumber Widya, [9] Adrial., (2006) Hubungan Faktor Lingkungan Fisik Rumah Terhadap Kejadian Tuberkulosis Paru BTA Positif di Kota Batam Kepulauan Riau Tahun 2005, Tesis, FKM-Universitas Indonesia. [10]Asriani., (2009) Hubungan Kondisi Rumah Dengan Kejadian Penyakit TB Paru di Wilayah Kerja Puskesmas Peruga Kota Bima Tahun 2009, Skripsi, FKM-Universitas Airlangga. [11] Budiyono F.X A., (2003) Faktor-Faktor Yang Berhubungan Dengan Kejadian Tuberkulosis Paru BTA Positif di Kota Jakarta Timur Tahun 2003, Tesis, FKM-Universitas Indonesia. [12] Ibupertiwi., (2004) Faktor-Faktor Yang BerhubungandenganKejadian TB ParupadaUsia 0-14 tahun di Kotamadya Jakarta Timurtahun 2004, Tesis, FKM-Universitas Indonesia. [13] Jaya Iman., (2000) StudiKasusKontrolFaktorResikoLingkunganTerhadapKejadian TB Paru BTA (+) di Kabupaten Aceh Barat tahun 1999, Tesis, FKM-Universitas Indonesia. [14]Kussularso., (1994) Studi Kasus Kontrol Faktor Resiko dari TB Paru di Kotamadya Surakarta tahun 1992, Tesis, FKM-Universitas Indonesia. [15] Leandro Gioacchino., (2005) Meta-analysis in Medical Research – The Handbook for the Understanding and Practice of Meta-Analysis, Oxford Blackwell Publishing. [16] Mahpudin A.H., (2006) Hubungan Faktor Lingkungan Fisik Rumah, Sosial Ekonomi dan Respon Biologis Terhadap Kejadian Tuberkulosis Paru BTA Positif pada Penduduk Dewasa di Indonesia (Analisis Data SPTBC dan Susenas 2004), Tesis, FKM-Universitas Indonesia. [17] Radhumi Okta Vinia., (2007) Faktor Yang MempengaruhiKejadianTuberkulosispadaAnakUmur 0-14 tahun di BalaiPengobatandanPemberantasanPenyakitParu (RS KarangTembok) Surabaya, Skripsi, FKMUniversitasAirlangga. [18] Romaita Nur., (2002) Faktor-Faktor Yang Berhubungan Dengan Terjadinya TB Paru pada Kelompok Usia Produktif di Wilayah Kerja Puskesmas Waru, Skripsi, FKM-UniversitasAirlangga [19] Silviana Ike., (2006) Hubungan Lingkungan Fisik Dalam Rumah dengan Kejadian TB Paru BTA (+) di Kabupaten Muaro Jambi Tahun 2005, Tesis, FKM-Universitas Indonesia. [20] Suarni Helda., (2009) Faktor Lingkungan Yang Berhubungan Dengan Kejadian Penyakit TB BTA Positif di Kecamatan Pancoran Mas Kota Depok Bulan Oktober tahun 2008-April tahun 2009, Tesis, FKM-Universitas Indonesia. [20] Suminar D.S., (2006) Studi Meta-Analisis Pretend Play dan Perkembangan Kognitif ,Insan, Volume 8, No. 1, April 2006 [21] Supriyadi., (2003) Hubungan Antara Kontak Serumah dan Faktor Lain Terhadap Kejadian TB Paru BTA (+) di Kota Banjarmasin tahun 2003, Tesis, FKM-Universitas Indonesia. [22] Supriyono D., (2003) Lingkungan Fisik Rumah Sebagai Faktor Risiko Terjadinya Penyakit TB Paru BTA Positif di Kecamatan Ciampea Kabupaten Bogor Tahun 2002, Tesis, FKM-Universitas Indonesia. [23] Yasiroh Nur., (2009) Pengetahuan, Sikap, Tindakan dan Lingkungan Rumah Penderita TB Paru di Wilayah Kerja Puskesmas Pegirian Tahun 2008-2009, Skripsi, FKM-UniversitasAirlangga.
42