INT J TUBERC LUNG DIS 6(8):666 671 2002 IUATLD Outcome of patients with tuberculosis who transfer between reporting units in Malawi S. Meijnen,* M. M. Weismuller,* N. J. M. Claessens,* J. H. Kwanjana, F. M. Salaniponi, A. D. Harries * Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands; National Tuberculosis Control Programme, Ministry of Health, Lilongwe, Malawi SUMMARY SETTING: All 43 non-private hospitals in Malawi which in 1999 registered and treated patients with tuberculosis (). OBJECTIVE: To determine the proportion of patients who transferred from one reporting unit to another and their treatment outcome, and to compare outcome results between the main register and the transfer-in register. DESIGN: Retrospective data collection, using the main register and transfer-in register, on all patients registered in Malawi in 1999. RESULTS: There were 24 908 patients, of whom 3249 (13%) in total were transfers. Significantly more patients transferred from mission hospitals (23%) compared with central (8%) or district (5%) hospitals (P 0.001). The date of transfer was recorded for 1406 patients, of whom 1170 (83%) transferred in the first 10 weeks. Respectively 45% and 58% of transfer patients had unknown outcomes in the main register and transferin register; these rates were significantly lower in smearpositive (P) patients. A total of 1357 patients were entered into transfer-in registers; 694 patients had matched names and/or registration numbers in both registers. Of the matched patients, 373 (54%) had similar treatment outcomes. CONCLUSION: It is common for patients to transfer between treatment units, but the quality of the data for patients who transfer is poor, and needs to be improved. KEY WORDS: tuberculosis; Malawi; transfer-out; transfer-in ONE OF THE CORNERSTONES of the World Health Organization s (WHO s) DOTS strategy for global tuberculosis () control is a standardised recording and reporting system. At the end of a course of treatment there are six possible outcomes: cure, treatment completed, failed, died, defaulted and transferred out. 1 A transfer out is a patient who transfers from one reporting unit ( registration centre) to another and for whom the treatment outcome is unknown. The proportion of smear-positive patients who are reported as transfers-out varies from one National Tuberculosis Programme to another. In sub-saharan Africa, for example, the transfer-out rates for patients with smear-positive tuberculosis (P) registered in 1998 ranged from 1% to 26%. 2 The Malawi National Tuberculosis Control Programme (NTP) has reported on the treatment outcome of smear-positive P cases since 1984, and on smear-negative P and extra- (EP) cases since 1998. During the 1990s, the percentage of new smear-positive P cases reported as transfersout remained fairly constant, at between 4% and 6% (source Malawi NTP). Reducing this rate of transfersout to as low a figure as possible is important for the Malawi NTP in its attempts to reach its cure rate target of 75%. (Malawi set its cure rate target at 75% because the high death rates due to human immunodeficiency virus [HIV] related disease meant it was impossible to reach the WHO s cure rate target of 85%.) Treatment outcomes for patients with all types of are recorded in the registers of each reporting unit. When a patient moves to another unit during the course of treatment, a series of actions should take place: a copy of the transfer form is posted to the new unit by the officer who first registered the patient; the patient also takes a copy of the transfer form to give to the officer at the new unit. At the new unit, the patient s details are entered into a transfer-in register. At the end of treatment, standardised outcome results are entered into the transfer-in register and are communicated back to the original registration unit for entry into the main register. If the system works at maximum efficiency, all patients who transfer to another reporting unit should appear in the transfer-in registers, and the outcome results should be communicated back to the initial registration unit. In order to find out more about patient transfers, we conducted a country-wide study of all patients Correspondence to: Professor A D Harries, c/o British High Commission, PO Box 30042, Lilongwe 3, Malawi. Fax: ( 265) 772 657. e-mail: adharries@malawi.net Article submitted 10 December 2001. Final version accepted 7 May 2002. [A version in French of this article is available from the IUATLD Secretariat in Paris.]
patient transfers and treatment outcome 667 who were registered with in 1999. The objectives of the study were to determine the proportion of patients who transferred from one reporting unit to another and the treatment outcome for transfer patients, and to compare outcome results between the main register and the transfer-in register. METHODS Setting In 1999, there were 43 non-private hospitals in Malawi that registered and treated patients, comprising three central and 22 district hospitals supported by the government, and 18 mission hospitals. There were also two private hospitals that registered and treated a small number of patients these hospitals were not included in the study. Study definitions A transfer was defined as a patient who transfers from one reporting unit to another, regardless of the eventual treatment outcome result. In three districts (Mulanje, Chikwawa and Machinga), where there are both district and mission hospitals together, there was an understanding between the officers that transfers between the district and mission hospitals within the district were not real transfers. Hence any district/mission transfers within these districts were not recorded in the registers as transfers. A transfer-out was defined as a patient who transfers and for whom the treatment outcome is unknown. Transfer and transfer-out data were collected from the main register. A transfer-in was defined as a patient transferring into a new reporting unit. Transfer-in data were collected from the transfer-in register. Data collection All 43 of the non-private hospitals that were registering patients in 1999 were visited between April and June 2001. Private hospitals were not visited because of the very low numbers of patients. The main register for 1999 was obtained at each hospital, and a record was made of the number and type of all patients registered between 1 January and 31 December 1999, and the number transferred (this information is recorded in the remarks column and/or the transfer column). Of patients who transferred, a record was made of the registration number, name, age, sex, type and category of, date and month of transfer, district to which the patient transferred and treatment outcome. The transfer-in register for 1999 and 2000 was obtained at each hospital, and all patients who transferred in with a registration number for the year 1999 were noted. For each patient, similar information to that obtained from the main register was recorded. Problems with data collection from the registers Registers In Lilongwe district, the transfer-in register consisted of two books, one of which was lost and could not be found. In Mangochi district, the transfer-in register only contained six patient entries, suggesting that either the register was not completely filled in or that there was also another book which had been lost. Patient information in the registers Some of the patient information in the registers, such as registration number, age, sex, type of, month of transfer and district of transfer, was not recorded, and hence there was incomplete information for some patients. In some transfer-in registers transfer-in registration numbers were incomplete, clearly wrong or not recorded at all. If the date of starting treatment was recorded and the year was 1999, the patient and the clinical details were entered into the study proforma. In all other cases, the patient and clinical details were not included. In some districts, the transfer in the remarks column/transfer column and the transfer out in the treatment outcome column had been deleted: this was usually when a treatment outcome had become known. On a case-by-case basis, working closely with the district officer, decisions were made in these situations about whether the patient actually transferred or not. Data analysis Data were collected into structured proformas. Two separate Excel spreadsheet files were developed for the transfer patients recorded in the main register and for the transfer-in patients in the transfer-in register. Excel files were used to compare and match patients between the two registers. Patients were considered to be matched if 1) the names and the registration number were the same, and 2) the names were the same even though the registration numbers differed, providing the age, sex and type of were similar. Patients were considered to be unmatched in any other circumstance. Data from the Excel spreadsheets were also entered into a series of Epi-Info files (version 6.04, CDC, Atlanta, GA). Proportions were compared using the 2 test, with differences at the 5% level regarded as significant. RESULTS Characteristics of transfer patients There were 24 908 patients registered with in the main register in 1999, of whom 2586 (10%) were transfers. Transfers included 1383 men, 1202 women and one patient whose sex was not recorded; the mean age (standard deviation [SD]) was 32 (14) years. The majority of patients (96%) had new.
668 The International Journal of Tuberculosis and Lung Disease Table 1 Patient transfers according to type of tuberculosis () and type of hospital Patients registered n Patients transferred Type of Smear-positive P 8 922 901 (10) Smear-negative P 10 372 1070 (10) Extra- 5 614 615 (11) Type of hospital Central 8 478 649 (8) District 10 481 555 (5) Mission 5 949 1382 (23) The number and proportion of patients who transferred according to type of and type of hospital are shown in Table 1. There were no differences in transfers according to type of, but significantly more patients who registered at mission hospitals transferred to another reporting unit compared with those registered at central or district hospitals (P 0.001). The date of transfer was recorded for 1406 (54%) patients. In patients where the date was known, the cumulative number and proportion of patients transferring were 435 (31%) in the first 2 weeks of treatment, 818 (58%) in the first 6 weeks, and 1170 (83%) in the first 10 weeks. Treatment outcome of transfer patients In the main register, treatment outcomes of transfer patients are shown in Table 2. The transfer-out rate (i.e., transfer with unknown outcome) was 45% for all patients, but was significantly lower for patients with smear-positive P compared with patients who had smear-negative P and EP (P 0.001). Treatment outcome of transfer-in patients In the transfer-in register, there were 1357 patients whose registration number was for the year 1999: 720 men, 635 women, and two whose sex was not recorded; their mean age (SD) was 33 (16) years. The date of transfer-in was recorded for 252 (19%) patients. In patients where the date was known, the cumulative number and proportion of patients transferring in were 76 (31%) in the first 2 weeks of treatment, 116 (46%) in the first 6 weeks, and 173 (69%) in the first 10 weeks. The treatment outcomes for transfer-in patients are shown in Table 3: 58% of all patients had an unknown outcome; this rate was significantly lower for patients with smear-positive P compared with those with smear-negative P and EP (P 0.001). Comparison between transfer and transfer-in patients The Figure illustrates the number of transfer patients in the main register who were matched and not matched when compared with the patients registered in the transfer-in register. There were 2586 transfer patients in the main register and 1357 in the transferin register. Thus, 1229 (48%) patients who transferred were not entered into any transfer-in register. Of the 2586 transfers in the main register, 694 (27%) were found in the transfer-in register, while 663 were found in transfer-in registers who were not recorded as transfers in the main register. Thus, the total number of patients who transferred during treatment according to the data collected from both the main register and the transfer-in register was 3249 (13% of all patients registered in 1999). Of 694 patients matched in both registers, 373 (54%) had a treatment outcome that was similar in each register, while 321 (46%) had an outcome that was different in each register (for example, a matched patient was recorded as treatment success in the main register and as died in the transfer-in register). Of those patients with similar treatment outcomes in the two registers, 196 (53%) had a known treatment outcome (for example, cured or died), and 177 (47%) had an unknown outcome. Of the 1357 transfer-in patients, the district from which the patient had transferred was known for 1295 (95%). Of these, 942 (73%) had transferred from a district within the same region, while 353 (27%) had transferred from a different region. Table 2 Patient transfers and treatment outcomes according to type of tuberculosis Treatment outcome All types of Patients who transferred in the main register with Smear-positive Smear-negative EP All outcomes 2586 901 1070 615 Treatment success* 1059 (41) 534 (59) 304 (29) 221 (36) Died 300 (12) 103 (11) 121 (11) 76 (12) Defaulted 56 (2) 8 (1) 34 (3) 14 (2) Failed 7 6 1 0 Transfer-out (i.e., unknown outcome) 1164 (45) 250 (29) 610 (57) 304 (50) * Treatment success sum of patients cured and those who completed treatment completed with no smears examined. One patient with smear-negative P was recorded as having positive smears towards the end of treatment. EP extra- tuberculosis.
patient transfers and treatment outcome 669 Table 3 Patient transfer-ins and treatment outcomes according to type of tuberculosis Treatment outcome All types of Patients in the transfer-in register with Smear-positive Smear-negative EP All outcomes* 1357 540 517 273 Treatment success 371 (27) 231 (43) 93 (18) 44 (16) Died 147 (11) 62 (11) 54 (10) 29 (11) Defaulted 25 (2) 9 (2) 11 (2) 5 (2) Failed 1 1 0 0 A second transfer 32 (2) 14 (3) 8 (2) 9 (3) Unknown outcome 781 (58) 223 (41) 351 (68) 186 (68) * Includes 27 patients where type of was not indicated in the register of these 3 had treatment success, 2 died, 1 was a second transfer and in 21 the outcome was unknown. Treatment success sum of patients cured and those who completed treatment with no smears examined. Second transfer to another district. EP extra- tuberculosis. DISCUSSION This country-wide study shows that if the number of transfers from the main register are added to the number of patients in transfer-in registers who could not be matched with any transfer name in the main register, the proportion of patients who transferred Figure Comparisons of transfers and transfer-ins between main register and transfer-in register. in 1999 was at least 13% of all registered cases. This has to be regarded as a minimum figure, for two reasons. First, some officers did not regard transfers between the district and mission hospital in the same district as transfers, whereas in the strict sense of the definition they are (i.e., patients are transferring from one reporting unit to another). Second, we may not have included some transfer patients because vital registration data were missing from both the main register and the transfer-in register. Transfer rates were highest among patients who initially registered at mission hospitals. Although antituberculosis treatment is free in these facilities, patients have to pay for hospital bed stay, food and other adjunctive therapies, and not surprisingly may decide to move to a government facility where all these amenities are free of charge. This would tie in with the data indicating that the majority of patients transferred early on during the course of treatment. In Malawi, a rifampicin-based regimen is given to smear-positive patients during the initial phase of treatment, and transfer with loss to follow-up at this stage is of concern because of the potential risk for developing drug-resistant. The proportion of patients who transferred was similar, regardless of the type of. However, unknown treatment outcomes in both the main register and the transfer-in register were significantly less in those with smear-positive P compared with patients with smear-negative P or EP. This probably reflects the importance given to patients with smear-positive P by NTP staff, and the general lack of attention paid to treatment outcomes in patients with other types of. Nearly half of the patients who were matched by name and/or registration number had a different treatment outcome recorded in the main register and the transfer-in register. This disparity is of concern. Possible explanations include officers 1) being informed incorrectly about data in the transferin registers, 2) incorrectly recording data coming from transfer-in registers, or 3) fabricating data.
670 The International Journal of Tuberculosis and Lung Disease We do not know whether these problems with patient transfers occur in other sub-saharan African countries, as we have not been able to find any previous references to the subject. However, in countries where the infrastructure is weak, communications are poor and health care worker numbers are small, we might expect similar problems to occur. What can be done to remedy the situation? First, there are a number of activities which could be carried out and improved upon within the offices themselves. officers need to ensure that transfer forms are fully completed, in clear hand-writing with correct vital registration data. There should be a sufficient supply of stamps so that the forms can actually be posted. A copy of these forms should be filed for each patient in a special transfer form folder. Information on patients transferring in must be entered correctly in the transferin register, and another folder kept in which to file these transfer-in forms. There is currently no column in the transfer-in register for recording date of transferin, and the registers therefore need to be amended. Second, officers must visit the health centres in their catchment area on a regular basis. The current study suggests that some transfers from health centres occur during the continuation phase, with the transfer information not being passed to the officer and consequently not being recorded in the main register. Similarly, some patients on transfer-in to a new district may go straight to a health centre and never report to the officer at the hospital; this information is also not passed on to the office. officers therefore need to train and supervise health centre staff on management of transfers, and during visits they should collect transfer forms and other vital data in the health centre registers for entering into the main register. Third, there is already a system of quarterly meetings between the regional officer and the district officers to peer review registers, discuss operational problems and share transfer-out and transfer-in data. This study suggests that the sharing of transfer data is not done to maximum efficiency. officers need to bring their register, the transfer-in register and the folders containing transfer forms to these meetings. Under the chairmanship of the regional officer, care and effort must be taken to ensure that all data are correctly shared. Finally, in their structured district supervisions, regional officers should include a check of the transfer-in registers and the folders containing transfer forms. Attention to these details may improve a system that is clearly not working properly. During the course of anti-tuberculosis treatment a substantial proportion of patients will move from one reporting unit to another, and it is important for the NTP to keep track of these movements. Better quality information in the long-term has to mean better quality care and better NTP performance. Acknowledgements We thank the officers who assisted in extracting patient registration and treatment outcome data from the main registers and transfer-in registers. We thank the Department for International Development (DFID), UK, the Norwegian Agency for Development Cooperation (NORAD) and the Royal Netherlands Tuberculosis Association (KNCV) for financial support. The study received the support of the Programme Steering Group and ethical approval from the Malawi Health Science Research Committee. References 1 World Health Organization. Treatment of tuberculosis. Guidelines for National Programmes. 2nd ed. WHO//97.220. Geneva: WHO, 1997. 2 WHO Report 2001. Global Tuberculosis Control. Communicable Diseases. WHO/CDS//2001.287. Geneva: WHO, 2001. RÉSUMÉ CONTEXTE : L ensemble des 43 hôpitaux non-privés du Malawi qui en 1999 ont enregistré et traité des patients atteints de tuberculose (). OBJECTIF : Déterminer la proportion de patients transférés d une unité à l autre et le résultat de leur traitement, et comparer les résultats finaux entre le registre principal de et le registre de l unité vers laquelle s est fait le transfert. SCHÉMA : Recueil rétrospectif des données, utilisant le registre principal de et le registre de l unité vers laquelle le transfert se faisait chez tous les patients enregistrés au Malawi en 1999. RÉSULTATS : Sur les 24 908 patients, il y a eu au total 3249 (13%) transferts. Le nombre de patients transférés est significativement plus élevé à partir des hôpitaux de mission (23%) par comparaison avec les hôpitaux centraux (8%) ou de district (5%) (P 0.001). La date du transfert a été enregistrée chez 1406 patients dont 1170 (83%) ont été transférés au cours des 10 premières semaines. Les résultats finaux dans le registre principal et dans le registre de transfert étaient inconnus respectivement pour 45% et 58% des patients transférés. Ces taux sont significativement moindres chez les patients tuberculeux pulmonaires à bacilloscopie positive. On a enregistré 1357 patients dans les registres de l unité vers laquelle le transfert avait lieu et pour 694 patients les noms ou les numéros d enregistrement étaient concordants dans les deux registres. Parmi les patients ayant fait l objet de comparaison, chez 373 (54%) les résultats du traitement étaient similaires. CONCLUSION : Il est courant que les patients soient transférés entre les unités de traitement, mais la qualité des données chez les patients transférés est piètre et doit être améliorée.
patient transfers and treatment outcome 671 RESUMEN MARCO DE REFERENCIA : La totalidad de los 43 hospitales no privados de Malawi, que registraron y trataron pacientes con tuberculosis () en 1999. OBJETIVO : Determinar la proporción de pacientes con trasladados de una unidad, que los había declarado, a otra y el resultado final de su tratamiento y comparar los resultados finales entre la unidad donde se realizó el registro principal (registro principal) de y el registro de la unidad donde fueron trasladados (registro de traslado). MÉTODO : Recolección retrospectiva de datos, utilizando el registro principal y el registro de traslado de todos los pacientes registrados en Malawi en 1999. RESULTADOS : El total de pacientes fue de 24.908, de los cuales 3.249 (13%) fueron trasladados. Una cantidad significativamente mayor de pacientes fueron trasladados a partir de los hospitales de misión (23%), en comparación con los hospitales centrales (8%) o distritales (5%) (P 0,001). Los datos del traslado fueron registrados para 1/.406 pacientes, de los cuales 1.170 (83%) fueron trasladados durante las primeras 10 semanas. Los resultados finales eran desconocidos para el 45% y el 58% de los pacientes en el registro principal y en el registro de traslado, respectivamente. Estas proporciones son significativamente menores para los pacientes con tuberculosis pulmonar con baciloscopia positiva. Un total de 1.357 pacientes fueron registrados en el registro de traslado y para 694 pacientes el nombre y/o el número de registro eran concordantes en ambos registros. De estos pacientes, 373 (54%) tenían resultados similares del tratamiento. CONCLUSIÓN : Los pacientes son trasladados con frecuencia entre las unidades de tratamiento, pero la calidad de los datos relativos a estos pacientes trasladados es deficiente y debe ser mejorada.