|Year : 2021 | Volume
| Issue : 3 | Page : 417-424
Feasibility, uptake and real-life challenges of a rural cervical and breast cancer screening program in Vellore, Tamil Nadu, South India
Shalini Jeyapaul, Anu M Oommen, Anne George Cherian, Tobey Ann Marcus, Thabitha Malini, Jasmin H Prasad, Kuryan George
Community Health Department, Christian Medical College, Vellore, Tamil Nadu, India
|Date of Submission||30-Mar-2019|
|Date of Decision||31-May-2019|
|Date of Acceptance||01-Jun-2019|
|Date of Web Publication||02-Nov-2020|
Anu M Oommen
Community Health Department, Christian Medical College, Vellore, Tamil Nadu
Source of Support: None, Conflict of Interest: None
Background: Early detection of breast and cervical cancer by organized screening has been found to reduce mortality rates in trials, but documentation of programme results and challenges is rarely done from non-trial settings. This study reports results of a population-based cancer control programme in a rural block in Vellore, Tamil Nadu, population size (116,085), targeting a population of 18,490 women aged 25–60 years, between November 2014 and March 2018.
Methods: Village-based health education sessions were conducted by social workers, using trained volunteers and health workers to motivate eligible women. Screening was done at a secondary level hospital, by trained general physicians using visual inspection with acetic acid and clinical breast examination, followed by colposcopy, radiological imaging (breast) and biopsy as required.
Results: A total of 8 volunteers and 17 health workers motivated women for 93 health education and screening sessions, in 46 out of 82 villages. While 1,890/18,490 (10.2 per cent) were screened for breast cancer, 1,783 (9.6 per cent) were screened for cervical cancer, with a yield of 3.4/1,000 for cervical pre-cancer/cancer. The main challenges were creating time for screening activities in a busy secondary hospital and difficulty in ensuring treatment completion of screen-detected cases.
Conclusions: Population-based cancer screening programs can be offered by secondary hospitals that also run primary care services, to increase screening rates. Clear referral systems need to be established, bearing in mind that social factors, especially poor family support, may pose a threat to treatment, in spite of easy availability of cure.
Keywords: Breast, cancer control program, cancer screening, cervix, visual inspection with acetic acid
Key Message A cancer screening program integrated with existing primary and secondary care services in a rural block in Vellore, Tamil Nadu, resulted in a 10% coverage of eligible women in the first three and a half years, and highlighted implementation challenges of screening and ensuring treatment completion.
|How to cite this article:|
Jeyapaul S, Oommen AM, Cherian AG, Marcus TA, Malini T, Prasad JH, George K. Feasibility, uptake and real-life challenges of a rural cervical and breast cancer screening program in Vellore, Tamil Nadu, South India. Indian J Cancer 2021;58:417-24
|How to cite this URL:|
Jeyapaul S, Oommen AM, Cherian AG, Marcus TA, Malini T, Prasad JH, George K. Feasibility, uptake and real-life challenges of a rural cervical and breast cancer screening program in Vellore, Tamil Nadu, South India. Indian J Cancer [serial online] 2021 [cited 2021 Oct 27];58:417-24. Available from: https://www.indianjcancer.com/text.asp?2021/58/3/417/299719
| » Introduction|| |
Breast and cervical cancer are the most common cancers among Indian women, with higher breast cancer reported from urban populations and cervical cancer from rural populations., Age-adjusted incidence rates of cervical cancer in developing countries in South-east Asia and Africa are at least three to four times higher in comparison with developed countries, where mortality rates are also 10 times lower. Worldwide, cervical cancer has become one of the most common cancers among women in developing countries with only 15 per cent occurring in developed countries. Poverty and poor screening practices have long served as factors preventing progress in terms of reducing premature deaths caused by cancer. Currently, the two main modes of prevention of cervical cancer are infection prevention through human papilloma virus vaccination and early detection through screening.
The fact that most cases of cancer in India are diagnosed at a late stage indicate poor awareness and accessibility to measures for prevention and screening., Although periodic Papanicolaou (Pap) smear and mammography are the gold standards for screening for cervical and breast cancers, respectively, in low-resource settings, mass screening has been found to be more feasible and effective, with visual inspection with acetic acid (VIA) and Clinical Breast Examination (CBE).,,
The National Family Health Survey (NFHS)-4 reported that only 25.3 per cent of urban women and 20.7 per cent of rural women had ever undergone examinations of the cervix, while only 11.7 per cent of urban women and 8.8 per cent of rural women aged 15–49 years had undergone breast examination. The current level of efforts towards cancer screening needs strengthening, especially in rural and remote locations where public health care facilities may be limited or overburdened with curative services. While effectiveness of successful trials of cancer screening in India has been documented, reports of challenges associated with real-life implementation of screening programmes are scarce. This paper reports the results obtained from a rural cancer screening programme under practical programme settings, in a rural block in Vellore, Tamil Nadu, South India.
| » Methods|| |
Kaniyambadi, a rural block in Vellore district, with a population of around 116,085, is the primary service area for the Community Heath and Development (CHAD) program, of a tertiary, not-for-profit health care institution in Vellore. This area is also served by a 140-bedded secondary hospital with outpatient and inpatient services, mainly for general medicine, obstetrics and gynecology, and pediatrics. Primary health care in the rural area is delivered through female health care workers called 'health aides', who cover a population of around six thousand each. Besides delivering primary care through mobile clinics, patients are also benefited by the strong referral pathway to the secondary hospital and if needed, to specialists at the concerned tertiary institution in Vellore. A health information system (HIS), which captures vital statistics, including causes of deaths, immunization status, chronic disease, and the like, is also in place since the 1980s. Causes of death are determined through a process which involves verbal autopsy by health workers, verified by physician's assessment of such reports. Further details of the CHAD program have been described earlier. This block is also served by four government primary health centres (PHCs), of which three currently offer opportunistic screening for gynecological cancers.
A cancer screening programme was started in August 2014, for women aged 25–60 years, through the CHAD programme. This report assesses the effectiveness of this population-based cancer control programme in the first three and a half years, between November 2014 (start of first screening clinic) and March 2018.
Setting up of the programme
The programme commenced with training of two groups who were involved in field level activities for promoting screening: 17 primary care female health workers (health aides), who were already part of the primary care team of the CHAD programme, as well as a few volunteers. Eight volunteers were selected from six villages (two of the larger villages had two volunteers per village), recruited through self-help groups. The volunteers were chosen based on the following criteria: married women, housewives or not in regular employment, age 25–40 years, educated to at least eighth standard, able to read and understand Tamil. The health workers were also asked to consult village representatives and choose villages that showed an interest in the programme for the first phase, with an aim to eventually cover the whole block.
Both health workers and volunteers attended a training session on cervical and breast cancer: burden, risk factors, need for screening, and methods for screening. Screening was also offered for the health workers and volunteers.
Post-graduate residents and junior consultants in community medicine (18 women doctors) were trained by a gynecologist in performing VIA examinations and recognizing abnormalities through interactive theoretical classes using visual aids (teaching slides) as well as practical demonstrations and assisted examinations in outpatient clinics. The processes are summarized in [Figure 1].
Field level activities
Each volunteer/health worker was given a list of women aged 25–60 years in their villages, who needed to be motivated and referred for the health education sessions. The health workers/volunteers organized a small group education session in each selected village. Health education regarding cancer of the breast and cervix—burden, symptoms and signs, and screening—was provided by either a doctor or trained medical social worker. The educational aids used were either flipcharts or powerpoint presentation using a portable projector, depending on the location, which varied from common meeting places in the village, to roadside sessions next to worksites. The women were also informed about the date and time for the screening camp to be organized in the secondary hospital. They were also asked to pass on the information regarding screening to other women in their village. On the day of the camp, the women were transported to the clinic along with the volunteer/health worker who had gathered them in the village.
For some villages, two or more clinics were arranged, depending on demand and number of eligible women, as it was not possible to screen more than 50 per day, due to limitations of personnel. It was decided to first complete a single round in all the villages, with further camps arranged in the future, in order to ensure that women could be screened once in three to five years.
Screening clinics were held once a week, for two hours in the afternoon, at the secondary hospital. The procedure was explained orally at the clinic, by the trained social worker, followed by recording of signatures for consent. One or two trained doctors (post-graduate trainees/consultant—community medicine) screened women by visual inspection with 3 per cent acetic acid and CBE, [Figure 1]. Self-breast examination was also taught to each woman. Women who had inconclusive VIA results were referred to a gynecologist for repeat testing soon after the clinic. All VIA positive women underwent colposcopy using a portable colposcope (Gynocular), which has previously been shown to have a high agreement with stationary colposcopy (kappa: 0.998). Biopsies were sent to the pathology department of the concerned tertiary institution. The health workers later followed up the results of all diagnostic tests and informed the patients of the results and doctor's recommendations.
Exclusion criteria for cervical screening were current menstruation (test was offered to be done a week later), hysterectomy or refusal for pelvic examination.
Besides VIA positive cases, others detected to have polyps were offered an excision biopsy, while vaginal or cervical infections were treated. Some women with unhealthy cervices also had examination using a colposcope, although they were VIA negative. However, there was no addition to the final yield of cervical cancer/pre-cancerous lesions from this group.
The Institutional Review Board and Ethics Committee of the concerned tertiary institution approved this study to evaluate the effectiveness of the screening programme (IRB Min No. 11408, 27 June 2018). Patients participating in the screening programme provided informed signed consent.
| » Results|| |
A total of 93 screening clinics were held between November 2014 and March 2018 for 46 out of 82 villages in the block. Of these 46 villages, six were covered by volunteers and 40 by health workers. In large villages (e.g., population >2,000), only a section of the village was selected, based on the interest of the villagers.
The number of women aged 25–60 years in the entire block was 31,469, of which 18,490 (58.8 per cent) residing in the 46 programme villages formed the target population for the first cycle of the programme (around three years). The number of women who attended the health education sessions was estimated to have been around 20 per se ssion, covering approximately 1,800 women (9.7 per cent of the target population). However, screening was not restricted to those who had attended the session, as information was passed on to other women who were allowed to attend screening, following a brief explanation in the clinic. Of the 1,891 who attended the screening clinic, only 1,191 (62.3 per cent) had also attended the village-based health education.
The attendance at the screening clinic was 1,891 (10.2 per cent of the target population) in the first three and a half years of the programme, [Table 1]. The mean age of the women attending the screening was 37.4 years (range: 25-60 years). Of these, 1,783 (94.3 per cent) were screened for cervical cancer, while 1,890 were screened for breast cancer (one was accidentally missed). Women attending the clinic who were currently menstruating, had hysterectomy or refused pelvic examination were not screened for cervical cancer (108, 5.7 per cent).
|Table 1: Results from the cervical cancer screening programme in rural Vellore|
Click here to view
Women with doubtful VIA results (44, suspicious white patches needing confirmation by the gynecologist) were referred to the gynecologist, of whom 32 women were VIA negative and 9 were confirmed as positive. Four women did not report to the gynecologist for confirmation of findings, because they were unwilling to wait till the end of the screening clinic. While 5.5 per cent of those screened were detected as VIA positive by the screening doctor, the final yield of cancer/pre-cancerous lesions was 3.37/1,000 [Table 1].
The woman with screen-detected Stage III B cancer was started on radiation at the radiotherapy unit of the referral centre and completed treatment successfully, subsidized through the referral system between the screening centre and the tertiary institution. However, the woman with carcinoma in situ initially refused treatment because of poor family support, until severe pain was experienced 15 months later, by when she was found to have Stage III B cancer and was given chemotherapy and radiation therapy.
The three women with cervical intraepithelial neoplasia grade 1 (CIN I) were followed up six months later and were found to be normal on colposcopy and are on regular follow-up, [Table 1]. The subject with CIN II underwent a loop electrosurgical excision procedure, was declared disease-free after the procedure and is on regular follow-up.
Another woman who had been diagnosed with cancer cervix Stage II B in 2013 (before the start of the current screening programme) and defaulted on treatment, also came to the screening camp in 2016, with abnormal bleeding. She was started on palliative radiotherapy for recurrence but defaulted again due to poor social support. After two years, she was briefly hospitalized for palliative care and provided home-based palliative care until her death.
Of the 22 women with breast lumps, 19 (86.4 per cent) underwent imaging (ultrasonography or mammography as indicated), with three women refusing to undergo imaging, citing lack of time, previous scan showing benign lesion and no family support. Of the five women who were advised aspiration/biopsy, four underwent the same and were found to have benign disease/fibroadenoma, while one refused. The overall yield of benign breast disease was 5.8/1,000, with no cases of breast cancer detected [Table 2].
|Table 2: Results from the breast cancer screening programme in rural Vellore|
Click here to view
Other practical challenges faced, besides diagnosis, or treatment default, included: need for training a new set of doctors every year; ensuring that women doctors were freed from other responsibilities by duty exchanges with male doctors; adding a new clinic to the hospital's weekly schedule, which already included other special clinics (antenatal, high-risk antenatal, immunization, diabetes, pediatric, leprosy); waiting time of two–three hours for screen positive patients to undergo colposcopy by the gynecologist (only one available at the secondary hospital on a given day); need for providing transportation to women coming from far off villages and reasons for non-attendance such as refusal to forfeit a day's daily wages and feeling of being well among asymptomatic women [Table 3].
|Table 3: Summary of challenges faced during implementation of the rural screening programme|
Click here to view
| » Discussion|| |
The National Program for Prevention and Control of Cancer, Diabetes, CVD and Stroke (NPCDCS) aims to provide opportunistic cervical and breast cancer screening through primary health care services, but detailed reports of the progress have not been published. This paper reports the feasibility and coverage of a programme conducted in a low-resource rural setting, where cytology-based cervical cancer screening and mammography for breast cancer screening are not feasible.
The RE-AIM (reach, effectiveness—adoption, implementation and maintenance) framework for evaluation of health promotion was used to shape the evaluation., The reach (R) of the programme in terms of exact numbers of women who actually received educational messages was not available. This is an important lesson to programme planners as capturing such information needs to be built into programme implementation. Although our educational sessions were estimated to have covered around only 10 per cent of the target population, more than a third of those who attended the clinic had come because of information obtained from others, having missed the educational session themselves. Thus, the women who attended the sessions spread the message to others, serving as peer educators, emphasizing the important role of educating at least some people in the community to become health promoters.
Effectiveness (E) of the programme was measured in terms of proportion of women who were screened. According to the NFHS-4, 21.7 per cent of urban and 24.4 per cent of rural women aged 15–49 years in the state of Tamil Nadu have undergone cervical examination, while 14 per cent and 16.9 per cent, respectively, have undergone breast examination. However, the proportion who had undergone such examination as part of cancer screening was not reported. This programme in Kaniyambadi, rural Vellore has been effective in screening around 10 per cent of the eligible target group of rural women over three and a half years. This was achieved using a minimal level of resources, based on feasibility of adding such a component to the existing primary and secondary health care activities of a health care institution that is primarily involved in antenatal and pediatric care and treatment of illnesses such as diabetes and hypertension. In three and a half years, the programme adoption (A) was 55 per cent of villages in the block covered at least partly, by one round of screening (setting level adoption).
A similar cancer screening programme is being carried out in KV Kuppam, another rural block in Vellore district, by the RUHSA (Rural Unit for Health and Social Affairs) department, Christian Medical College Vellore. The processes in that programme are slightly different as screening is carried out by trained nurses at designated primary care facilities in the villages. This was not considered feasible for a programme that delivers primary care through mobile clinics. The proximity of the screening site to the population could partly explain the higher rate of screening obtained in the programme from KV Kuppam (16 per cent in three years), compared to 10.1 per cent in this report. The more intensive educational efforts undertaken as part of the programme in KV Kuppam, such as training around 1,000 peer educators, would have also contributed to the greater numbers screened.
Despite its constraints, this programme evaluation proves that even with minimal extra inputs, especially in situ ations where there are staff constraints limiting additional responsibilities, it is possible to have fair results in community-wide screening programmes in similar settings. The programme needs to be supported by a trusted health care system with adequate facilities at a secondary level as well as a strong referral network.
The coverage of the target population from these rural community-based programmes (16 per cent in KV Kuppam, 10.2 per cent in Kaniyambadi) was far lower than would be obtained in a clinical trial setting, where resource-intensive efforts are employed to recruit and follow up subjects. In the clinical trial of breast and cervical cancer screening among the urban poor in Mumbai, 84 per cent participated in at least one round of VIA screening and 93 per cent in CBE. The coverage achieved in these community-based screening programmes from Vellore district was also far lower than the 55 per cent participation rate seen in Australia, through national breast and cervical cancer screening programmes and may be related to barriers faced due to illiteracy, fears related to screening, poor understanding of risk, and the like The fact that there was no yield of breast cancer cases in this rural programme is not surprising, as breast cancer in India is more common in urban areas, although the incidence is rising throughout the country., Use of mammography for screening could have improved yield but was not a feasible option for this population-based screening programme. The yield of cervical cancer/pre-cancerous lesions in this study was 3.37/1,000, similar to the programme from rural KV Kuppam (2.8/1,000) and the clinical trial in Mumbai (2/1,000).,
Implementation (I) evaluation of the programme revealed the difficulty in ensuring that all those who were detected by screening could complete the further processes. Of the six women with pre-cancerous/cancerous lesions, one woman did not agree for treatment, and returned with disease progression 15 months later, which was fortunately still treatable. Also, four women did not complete the cervical screening process (confirmation by the gynecologist and colposcopy) and four women did not come for imaging/tissue testing for breast lumps. Such difficulties in follow up were also reported by the KV Kuppam programme.
Maintenance (M) of the programme at the setting level depends on availability of inputs on a long-term basis. While the few volunteers who participated in the programme were motivated, the fact that only eight volunteers were obtained from a block with 1,20,000 persons shows the difficulty in obtaining people from rural areas who are able to spare time for purely voluntary work. This difficulty led us to realize that existing health workers will also need to be involved and a pure volunteer-driven approach is not feasible in this rural setting, where working for a livelihood is the obvious limiting factor for time contributions. The sustainability of the programme was tested when the three-year pilot period ended, and the department had to decide on whether to continue the services and which decision was made based on the fact that the programme had become an integral part of the health care delivery by the end of three years.
Maintenance of the practice of periodic cancer screening visits by individuals who have participated in one round of screening was not in the scope of this initial evaluation.
Limitations and challenges
Villages were chosen for the initial phase of this programme based on interest of the community and not based on randomized processes. As there was only one health worker/volunteer for a population of around 6,000, only a section of the village was chosen for larger villages, implying that all 46 villages were not reached completely.
As this was not a clinical trial, no rigorous measures were taken to ensure that all those who received education attended the screening camp, the decision being purely taken by the participants. Also, in order to promote maximal utilization of the services, those who did not attend the health education session were also allowed to come for the screening and were briefly explained the procedure in the clinic. This, however, served to improve clinic attendance, as a third of those who attended the screening had not attended the health education session in the village but had heard of the programme from other women.
Although clinics were planned every week, sometimes they had to be restricted to one or two per month, due to shortage of staff, leave and festival seasons, and so on, leading to fluctuations in numbers screened, a problem that has also been seen elsewhere. Using trained nurses or other health workers as done in clinical trials, NPCDCS and the rural KV Kuppam program,,,, was not possible due to shortage of nurses and lack of time for such additional responsibilities. Increasing availability of trained personnel who can replace or add on to the pool of doctors needs to be considered to enable more frequent screening clinics (scheduled around two per month in this project), as busy health centres have many other competing interests. Unlike in stringent trial settings, there was no objective evaluation of skills of the trained doctors, which is a limitation, that needs to be considered for the future.
Screening was held at the secondary hospital located in the block, as the concerned primary health care programme did not have fixed facility sites. This necessitated provision of transport facilities to the clinic whenever possible, as the villages were spread over a radius of 25 km from the secondary centre.
Data on proportion of women who did not attend the screening clinics because they had already been screened elsewhere (e.g., government health system) was also not documented. Although the coverage was only around 10 per cent in the first round of this programme, it can be reasonably expected that the group of women who attended the screening clinics may not have been screened in the absence of this programme. Ultimate proof of effectiveness of a cancer screening programme will be reduction in mortality due to the concerned cancers. As systems to capture causes of death are already in place in the study area, such an evaluation will also be done for long-term evaluation. Ensuring follow up of those with screen-detected cancers, enabling completion of treatment is essential to reduce mortality.
| » Conclusions|| |
This report documents the real-life implementation challenges, feasibility and utilization of a rural population-based cancer screening programme, run by a community health care organization, with a long history of health and development activities in the community. Although the coverage was lower than expected in this initial phase, the programme has been shown to be feasible and sustainable for implementation by a secondary care hospital with primary care outreach services and with further improvements, it can potentially impact the lives of many more rural women. As government-based screening was also available through the NPCDCS, this programme provided an additional avenue for screening. Low intensity population-based disease prevention programmes covering a large percentage of the target population may be more sustainable than high-intensity, resource-intensive interventions.
Unlike results of effectiveness and implementation of trials of screening which are readily available, reports of evaluations of ongoing health programmes are far less in developing countries. This report is one such attempt, to improve evidence from implementation science research in India. Future plans for this programme include covering the remaining 36 villages for the first round, ensuring periodicity of surveys for repeat screening and incorporation of the screening results into the existing HIS, with a view to maintaining a population-based cancer registry for cervical and breast cancer. Barriers leading to poor uptake of screening programmes also needs further documentation with efforts to modify programmes to reduce these. Stringent follow up of screen-detected patients with maximal efforts to ensure compliance to further diagnostic tests and treatment are additional measures needed to improve effectiveness of screening programmes.,,
Similar population-based screening strategies if undertaken by non-public sector secondary level health care settings and community medicine departments of medical colleges, can contribute to increasing coverage for cancer screening, especially in rural and remote areas in the country, supplementing opportunistic screening already available through PHCs.
Primary health care team and post-graduate residents from the Community Health Department, Christian Medical College Vellore. We thank Mr. Hugh Skeil, Development Office, Christian Medical College Vellore for his contributions towards project management and his valuable suggestions.
Financial support and sponsorship
Christian Medical College, Vellore.
Conflicts of interest
There are no conflicts of interest.
| » References|| |
Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2018;68:394–424.
Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer statistics. CA Cancer J Clin 2011;61:69–90.
Sankaranarayanan R, Esmy PO, Rajkumar R, Muwonge R, Swaminathan R, Shanthakumari S, et al.
Effect of visual screening on cervical cancer incidence and mortality in Tamil Nadu, India: A cluster-randomised trial. Lancet 2007;370:398–406.
Shastri SS, Mittra I, Mishra GA, Gupta S, Dikshit R, Singh S, et al.
Effect of VIA screening by primary health workers: Randomized controlled study in Mumbai. India. J Natl Cancer Inst 2014;106:dju009.
Mittra I, Mishra GA, Singh S, Aranke S, Notani P, Badwe R, et al.
A cluster randomized, controlled trial of breast and cervix cancer screening in Mumbai, India: Methodology and interim results after three rounds of screening. Int J Cancer 2010;126:976–84.
Isaac R, Finkel M, Oliver I, Annie IK, Prashanth HR, Subhashini J, et al
. Translating evidence into practice in low-resource settings: Cervical cancer screening tests are only part of the solution in rural India. Asian Pac J Cancer Prev 2012;13:4169–72.
Mohan VR, Muliyil J. Mortality patterns and the effect of socioeconomic factors on mortality in rural Tamil Nadu, south India: A community-based cohort study. Trans R Soc Trop Med Hyg 2009;103:801–6.
Nessa A, Wistrand C, Begum SA, Thuresson M, Shemer I, Thorsell M, et al.
Evaluation of stationary colposcope and the Gynocular, by the Swede score systematic colposcopic system in VIA positive women: A crossover randomized trial. Int J Gynecol Cancer 2014;24:339–45.
Glasgow RE, Vogt TM, Boles SM. Evaluating the public health impact of health promotion interventions: The RE-AIM framework. Am J Public Health 1999;89:1322–7.
Gaglio B, Shoup JA, Glasgow RE. The RE-AIM framework: A systematic review of use over time. Am J Public Health 2013;103:e38–46.
Aziz Z, Absetz P, Oldroyd J, Pronk NP, Oldenburg B. A systematic review of real-world diabetes prevention programs: Learnings from the last 15 years. Implement Sci 2015;10:172.
Krishnan S, Madsen E, Porterfield D, Varghese B. Advancing cervical cancer prevention in India: Implementation science priorities. Oncologist 2013;18(Suppl):13–25.
Sankaranarayanan R, Budukh AM, Rajkumar R. Effective screening programmes for cervical cancer in low- and middle-income developing countries. Bull World Health Organ 2001;79:954–62.
[Table 1], [Table 2], [Table 3]