Maternal and child health services in rural Nepal: does access or quality matter more?

Transcription

1 HEALTH POLICY AND PLANNING; 15(2): Oxford University Press 2000 Maternal and child health services in rural Nepal: does access or quality matter more? LAXMI BILAS ACHARYA 1 AND JOHN CLELAND 2 1 Central Department of Population Studies, Tribhuvan University, Kathmandu, Nepal, and 2 Centre for Population Studies, London School of Hygiene and Tropical Medicine, London, UK This study seeks to establish the relative importance of service access and quality on utilization of preventive health services in the western and middle-western Hill region of Nepal. Access was measured in terms of travel time to the nearest health post and coverage by outreach workers. The quality of static services was defined in structural terms: physical infrastructure, number of staff, availability of drugs and holding of special maternal and child health clinics. The initial analysis showed that no single indicator of quality was of overriding importance and therefore an overall quality index was constructed. After adjustment for access and for socioeconomic characteristics of families and communities, a very pronounced relationship between overall structural quality of the nearest health post and service uptake persisted. The adjusted odds of using some form of antenatal service were 6.6 times higher in the catchment areas of high quality posts than in areas served by low quality posts. The corresponding figure for receipt of BCG vaccination is 8.1. By comparison, the effects of travel time to the nearest health post are modest. Uptake of services is about twice as high when there is a health post in the community. Regular monthly visits by outreach workers also had a marked effect on service utilization. These results suggest that investment in the quality of health posts is more important than further increases in their number and that a further expansion of outreach services is a priority. Background Modern health services have a short history in Nepal. Until the early 1960s health care was restricted to a few urban hospitals and a few rural dispensaries. The Ministry of Health was established in 1956 and its first priority was to control communicable diseases. Four major vertical projects were initiated, namely malaria eradication (in 1958), leprosy and tuberculosis control (in 1965), and smallpox eradication (in 1967). Following the eradication of smallpox, this project became the expanded programme on immunization in The Ministry underwent a major reorganization in 1987, when all vertical projects were integrated under the Public Health Division. Simultaneously, five regional health directorates and 75 district public health offices (DPHOs) were established. At present, curative services are provided through central, regional, zonal and district hospitals, while primary health care and other services, previously provided by the vertical projects, are delivered through health posts. In 1991, there were 816 health posts, which are managed by DPHOs. Each health post has a sanctioned staff of five, though a minority have a staff of only three professionals. In addition, there are 18 health centres, with a larger complement of staff. Under the 1991 Health Policy it was decided to create a new tier of sub-health posts each staffed by a female maternal and child health (MCH) worker, an auxiliary field worker and a village health worker. However, at the time of this study the scheme had been implemented in only a minority of areas. Outreach services are delivered by village health workers and female community health volunteers. Village health workers are multipurpose field-based staff who are trained for 90 days in primary health care, including immunization. Although there is no sex discrimination in their recruitment, at present over 80% are male (UNICEF/HMG Nepal 1992). They conduct household visits, provide basic treatment and medicines, keep a record of pregnant women and all children under five, provide tetanus toxoid injections for pregnant women, make referrals when necessary and provide health education. They are supposed to motivate communities for family planning and teach the importance of child spacing, oral rehydration therapy and immunization. Village health workers are government employees, are based at health posts and receive a salary from DPHOs. Each worker is expected to spend 5 days per month at the health post. The remaining days of the month are spent on home visits and delivery of immunization services and tetanus toxoid injections to pregnant women in villages. In 1991 there were about 4000 such workers, each serving an average population of 5000 (Figure 1). 1 The female community health volunteers are mainly involved in motivational aspects of health provision, delivery of firstaid services, and resupply of contraceptive pills and condoms within their ward of residence. Working under the supervision of village health workers and health post staff, they are expected to promote the utilization of services and encourage the adoption of preventive health practices through mothers groups. Married women able to read and write are eligible for

2 224 Laxmi Bilas Acharya and John Cleland Figure 1. Health services at district level Family Planning and Health Survey (NFFHS), that collected household-level data on service utilization, as well as detailed demographic and socioeconomic data (NIV 1993). Surprisingly, this survey failed to collect information on access to, or quality of, services, thus severely reducing the practical utility of the results. Accordingly, a separate supplementary community survey was designed and conducted at the end of 1994 to remedy this omission. This survey collected limited information on access to and quality of services for 237 NFFHS sample clusters in the western and middle-western Hill region of Nepal from the records of DPHOs. More detailed information on access and quality was collected by visiting a subsample of 56 clusters, 28 health posts and 42 community health volunteers. These 56 clusters were selected using the criterion that the cluster should have a health post within three hours walking distance, the assumption being that quality may be a more important influence on service uptake when physical access is reasonable. In 1994, approximately 70% of people in the western and middle-western Hill region lived within about 3 hours travelling time of a health post. Since 1994, the number of health posts has been expanded and physical access has improved. Thus the results based on the sub-sample apply to a large majority, but not all, of the population of this sub-region of Nepal. A check of the reliability of data collected in both surveys shows that the quality of data is good (Acharya 1997). the post. They are given 24 days of training at the DPHO. Community health volunteers are not paid for their work. However, after initial training, they receive a fixed amount of money to buy first-aid medicines and non-clinical contraceptives, such as pills and condoms, which they can sell and thereby make some money. Such financial help is given once only. As of June 1991, community health volunteers had been trained, compared with the national target to train , which would represent a ratio of one volunteer per ward (UNICEF/HMG Nepal 1992). Study objectives The objectives of the study were straightforward. First we wished to assess the effect of physical access to services on their utilization. Access to health centres is a key issue in the Hill region of Nepal, where travel time has to be measured in hours or even days rather than minutes because of the topography and because most travel is on foot. We also wished to assess the effect of service quality on uptake. As in many very poor countries, rural health services in Nepal are beset with numerous deficiencies that affect in obvious ways the quality of services on offer. For instance, shortages of trained staff, equipment and drugs are common. There seemed little point, therefore, in estimating the influence of access without also taking into account the influence of quality. Indeed, the key research question that is addressed in this paper may be stated as follows: Does access or quality matter more in terms of people s willingness and ability to use services? Data sources The idea for the study stemmed from the existence of a large nationally representative survey, the 1991 Nepal Fertility, An analysis file was prepared by linking the NFFHS and 1994 community survey data. This paper is based on the data from the sub-sample, which includes information on 592 children under five from 56 clusters and on 28 health posts. 2 Selection and measurement of variables Receipt of antenatal care and BCG immunization were selected as the two measures of service utilization for the study. Both are coded as binary variables and are defined as detailed below. Outcome variables Antenatal care All mothers in the 1991 NFFHS were asked if they had received antenatal services, and if so, whom they had seen during the pregnancy of each child born in the last 5 years. They were also asked whether they received a tetanus toxoid injection during the pregnancy of each child. All those who received some help or advice, whether from a doctor, a nurse or an auxiliary midwife, or who were injected for tetanus during the pregnancy of the index child were categorized as users of some antenatal services. All other women who did not use any of these services were categorized as non-users. About 40% of mothers had used such services (38% tetanus toxoid; 15% other antenatal services) in the study sample. BCG vaccination In the NFFHS mothers reported whether or not each child under five had received BCG, polio, DPT and measles vaccination. For the present analysis BCG vaccination was chosen

3 Utilization of MCH services 225 as a suitable summary indicator of use of immunization services because of its high correlation with immunization against other childhood diseases. Furthermore, mothers recall of BCG vaccination is likely to be better than for other vaccinations because it leaves a scar on the arm of the child. Slightly over 60% of children included in the analysis had received BCG vaccination. Access to health services Access to the nearest health post was measured in terms of travel time from the centre of each cluster, based on normal mode of transport, typically walking or bus. Proximity to the nearest medical store was assessed in the same way. The presence of village health workers was defined in terms of frequency of visits to the village, as reported by key informants. Finally an index of the activeness of community health volunteers was constructed from interviews with volunteers. Information on the numbers of mothers meetings held, average attendance at such meetings, stock of medicines and so on was used to classify volunteers as active and less active. It should be noted that these measures of outreach activity relate to the situation in 1994, whereas the service utilization data were obtained in This period spans Nepal s transition to democracy and appreciable changes to the structure of local government. Reforms to primary health care were also initiated but, as mentioned earlier, implementation was still at an early stage in Changes between 1991 and 1994 in provision of services were probably minor but nevertheless may dilute the estimated relationship between access and uptake. A total of 12 of the 56 clusters had a health post in the village. Medical stores are somewhat less numerous. Only eight clusters had a medical store. About half (27) of clusters received regular monthly visits from village health workers but 21 were not covered at all by this type of outreach service. Only 19 clusters had a female health volunteer, of whom eight were classified as active. Quality of static services In this study, no attempt was made to measure all dimensions of service quality. Rather, attention was restricted to structural aspects of quality, such as availability of trained staff, equipment, supplies and facilities. Each of the 28 health posts was visited, staff were interviewed, and an inventory of drugs and equipment was taken. From these data, seven indicators of quality were derived: a physical facility index (based on number of rooms, the presence of a separate examination room, the existence of staff quarters, functional electricity and piped water, and nature of toilet facilities); number of professional staff; frequency of supervision by District staff; the holding of separate MCH clinics; staff involvement in private clinics; 3 availability of drugs on the day of visit; and availability of equipment. Again it should be noted that some of these measures of quality relate to 1994 rather than Preliminary analysis indicated a high correlation between the availability of equipment and drugs and, accordingly, availability of equipment was dropped. Eighteen health posts had supplies of half or more of the recommended drugs, but only eight had a full complement of professional staff and an identical number received monthly supervisory visits from DPHOs. Staff involvement in private clinics is common (17 out of 28) though not universal. Only 12 health posts ran separate MCH clinics. As expected, the structural quality of these 28 health posts is related to their accessibility to DPHOs. Health posts that are located in valleys rather than hill sides, and that are linked to headquarters by road, tend to have better physical facilities and drug supplies and to be more frequently supervised than posts that are less well connected (Table 1). Accessibility by road appears to be more important in this regard than travelling time. Clearly any analysis of the effect of quality of service uptake needs to take into account the fact that poor quality posts tend to be inaccessible and remote. Table 1. Correlation coefficients between proximity to district headquarters and location of health post and quality indicators Quality indicators Proximity to DPHO Location of health post c Mode of Travel time b transport a Physical facilities index * Number of professional staff MCH clinics held * Supervision of health post * * * Private clinics held by staff Supply of drugs * * p < a Variable was coded: foot = 0, mixed = 1, bus/minibus = 2. b Variable was measured in hours. c Variable coded: hill/mountain = 0, valley = 1. DPHO = district public health office; MCH = maternal and child health.

4 226 Laxmi Bilas Acharya and John Cleland Table 2. Correlation coefficients between six quality indicators of health posts and nature of outreach services (Pearson s coefficient) Quality indicators of health posts Frequency of visits by Activeness index of village health workers community health volunteers Physical facilities index * Number of professional staff * MCH clinics held * Supervision by DPHO * * Private clinics held Supply of drugs * Total cases * p < DPHO = district public health office. Is there a similar link between quality of static health posts and the activities of outreach workers in their catchment area? Table 2 shows correlation coefficients between the six quality indicators and two outreach variables: frequency of visits by village health workers and the activeness index for community health volunteers. Though most of the relationships are in the expected direction, the only consistently strong association concerns health post supervision. In areas where the health post is regularly supervised by DPHO staff, outreach activities of both types appear to be more vibrant. Because of this close link, the decision was taken not to include both types of variable in the same explanatory model. Exploratory analysis was also undertaken to assess whether clusters that were close to the nearest health post were better served by outreach workers than other clusters. Results (not shown) suggest that there is the opposite tendency; clusters close to a health post were less likely to receive visits from village health workers and less likely to have a volunteer than other clusters further away. This pattern reflects a deliberate policy to focus outreach activity on communities that have poor access to static services. Individual, household and community characteristics Uptake of services is likely to depend not only on access to services of reasonable quality but on a host of individual, household and community characteristics that may condition demand for modern health services. From the information collected in the NFFHS, the following variables were selected: mother s age; number of living children; number of births in the preceding 5 years; birth order; education of mother and husband; occupation of mother and husband; the socioeconomic status of the household (an index based on radio ownership, type of toilet and nature of water supply); and ethnicity. Two community-based factors were measured from observation and key informant interviews. First, an ethnicity variable was defined as follows: if a single ethnic group comprised more than 60% of population, the cluster was classified as ethnically homogenous. Such clusters were then subdivided into two groups: those dominated by higher Hindu castes, Brahmins or others (21 clusters), and those dominated by Tibeto-Mongoloid groups, Magars, Gurungs or Tamangs (23 clusters). The remaining 12 clusters were classified as ethnically mixed. Secondly, a cluster remoteness index was defined in terms of proximity to district headquarters, primary school, post-office and market centre, as well as the presence of electricity and irrigation. On the basis of this information, 10 clusters were classified as highly remote, 16 as less remote and the remaining as intermediate. Analytical method In order to estimate the effects of service access and quality on uptake of services, multilevel logistic regression was used. A logistic approach was needed because both outcome variables are dichotomous, and a multi-level approach was required because data are available at three levels of aggregation: health post, cluster and household or individuals (Aitkin and Longford 1986; Aitken et al. 1989; Woodhouse et al. 1995). As noted earlier, the need for a fourth level, children, was circumvented by selecting only one child under five per woman. The first step in the analysis was to examine unadjusted associations between predictor and outcome variables. In the subsequent construction of multivariate models, only variables showing significant unadjusted associations at a 10% level were retained, except for variables of key interest which were retained automatically. Results are shown in terms of odds ratios, and Wald statistics are used to test the significance of effects. First-order interactions were tested. Results Two factors (access to medical store and the holding of private clinics) were omitted from the multivariate models because of their lack of association with outcome variables. A third factor (supervision of health posts) was also dropped because of its close association with outreach activities. Most of the remaining indicators of access and quality have statistically significant unadjusted associations with the service use outcomes (Table 3). In terms of net or adjusted results, travelling time to the nearest health post has a significant effect on both outcomes. Compared to communities that are 2 3 hours away from a health post, use of both antenatal and child immunization services is higher when the health post is located within the community. The frequency of visits to a community by village health workers has a significant net effect on receipt of antenatal care, but only when visits are on a regular monthly basis. For BCG vaccination, however, such visits have no significant net effect. With regard to community

5 Utilization of MCH services 227 Table 3. Odds ratios of service utilization by the categories of selected variables Variables No. of Antenatal care BCG vaccination children Unadjusted Adjusted a odds Unadjusted Adjusted a odds odds ratio ratio (95% C.I.) odds ratio ratio (95% C.I.) Travel time to health post In cluster * 2.16* ( ) ( ) <2 hours * 0.91 ( ) ( ) 2+ hours Visits by VHW None Not every month ( ) ( ) Every month * 2.05* ( ) ( ) Activeness of CHV None Less active * ( ) ( ) Active ( ) ( ) Index of physical facilities Low Medium * 1.55 ( ) 3.44* 2.56* ( ) High * 1.05 ( ) 4.08* 1.74 ( ) Professional staff Less than sanctioned As sanctioned * 2.45* ( ) 2.51* MCH clinic No Yes * 1.14 ( ) 2.77* 2.34* ( ) Supply of drugs <50% % * 1.71 ( ) 2.11 Variance at cluster level * Health post level log likelihood * p < 0.05;, not included in the model. a Adjusted for: births in last 5 years, respondent s education, husband s education, and socioeconomic index. VHW = village health worker; CHV = community health volunteer; MCH = maternal and child health. health volunteers, BCG coverage tends to be higher in villages with a volunteer than in other villages though the difference is not statistically significant. There is also a positive effect on receipt of antenatal care. Unexpectedly, however, antenatal coverage is significantly higher in communities where the activeness of volunteers was assessed as medium but not where it was assessed as high. The quality of health post services is represented by four factors: physical facilities, number of professional staff, holding of MCH clinics and supply of drugs. The pattern of results differs between the two outcome variables. Thus, number of professional staff and drug availability are significantly related to antenatal care but not to child immunization. Conversely the holding of separate MCH clinics has a significant effect on BCG uptake but not on receipt of antenatal care. The physical infrastructure of the health post is significantly related to both forms of service utilization though the effect is not monotonic. Thus far in the analysis, the quality of static health services has been represented by its components. As no single component of quality has emerged as an overridingly important or decisive influence on service uptake, it was decided to reassess the quality utilization relationship by forming one single index of quality. Such a single index also makes sound substantive sense, because the effects on utilization of the different components of quality are likely to be mutually reinforcing. For instance, a full complement of professional staff in a health post may not by itself attract potential users if the physical infrastructure is decrepit or drug supply poor, and vice versa. Overall quality was defined as follows. Health posts having a full complement of professionals, with at least 50% of recommended drugs in stock, that hold separate MCH clinics and have a reasonable physical infrastructure (at least two of the six basic facilities discussed earlier) were classified as high quality. Health posts falling short of all these criteria were classified as low quality. The residue forms an intermediate group. According to this classification scheme, six posts fall into the high category, five into the low category and the majority (17) are intermediate. The unadjusted and adjusted effects of overall quality on service utilization are shown in Table 4. It is immediately clear that the effects of overall quality are very strong. The adjusted odds of use of antenatal care and BCG vaccination are 6.58 and 8.13 times higher, respectively, in the catchment areas of

6 228 Laxmi Bilas Acharya and John Cleland Table 4. Effect of overall quality of health posts on service utilization Quality of No. of Antenatal care BCG vaccination health post children Unadjusted Adjusted odds Unadjusted Adjusted odds odds ratio ratio (95% C.I.) odds ratio ratio (95% C.I.) Low Medium * 2.90* ( ) 6.81* 6.6* ( ) High * 6.60* ( ) 10.22* 8.1* ( ) * p < high quality posts than the odds in low quality ones. Furthermore, the odds of service utilization are also significantly higher in the intermediate than in the low quality category. Conclusions and policy implications In order to clarify the implications of these results for primary health care posts in Nepal, multilevel regression models were used to estimate the predicted probabilities of services use by a typical Nepalese family under varying conditions of access and service quality. 4 With the values of the characteristics of the typical family fixed, best fitting models were used to predict service uptake under different combinations of access and quality. The results are shown in Table 5. In each of the two panels, the first three rows show the estimated effects on utilization of different combinations of access to, and quality of, health posts in the absence of outreach activity. The fourth shows the estimated effects of outreach coverage in clusters where health post quality is rated to be low. Focusing attention initially on the first three rows of each panel, it can be seen immediately that, in the absence of outreach services, health post quality exerts a much more pronounced influence on service utilization than access. Antenatal coverage is four to six times higher in the catchment areas of high quality health posts than for low quality health posts. The relationship between quality and immunization is less striking but nevertheless appreciable. Immunization coverage is three to four times higher among families whose nearest health post is a high quality one than among those who have a poor quality post. The study thus strongly supports the conclusions of small scale enquiries that quality improvement is a priority (Baker et al. 1994; Freedman 1996). By comparison the effects of access are modest. In terms of antenatal care, including receipt of tetanus toxoid, utilization is nearly twice as high when there is a health post in the cluster than when there is no health post. But when there is no health post in the cluster, travel time is unrelated to service uptake. For child immunization, access to a health post makes even less difference. 5 Table 5 shows the estimated levels of coverage when health post quality is low but outreach activity is high (i.e. village health workers make regular visits and there is an active community health volunteer). The effect of outreach workers on uptake of antenatal services is pronounced. Coverage is four to five times higher in the presence of outreach activities than in their absence. This effect is almost as large as the effect of a change from low to high quality health posts. For child immunization, the relationship between outreach activities and coverage is less strong. Thus it appears that the provision of community-based services may be more decisive for women s than for children s health care. The conclusions of the study with regard to the access quality trade-offs are clearcut. The evidence strongly suggests that basic improvement to health post quality is a more important Table 5. Projected probabilities of using health services in the western and middle-western Hill region of Nepal (sub-sample) Access to VHW and CHV Health post Travel time to nearest health post quality 2 3 h <2 h in cluster Antenatal services No outreach services low No outreach services medium No outreach services high Monthly visits by VHW and active CHV in village low BCG vaccination No outreach services low No outreach services medium No outreach services high Monthly visits by VHW and active CHV in village low VHW = village health worker; CHV = community health volunteer.

7 Utilization of MCH services 229 priority than further increases in the number of health posts. Nearly one-fifth (18%) of health posts were judged to be of poor quality. They had less than their full complement of staff; less than half of recommended drugs were in stock; there was no piped water or electricity; and separate MCH clinics were not held. Under these circumstances, it is not surprising that service utilization tends to be low. The study has also demonstrated an important contribution of village health workers to the use of MCH services, particularly for women. Out of 56 clusters, only 27 (48%) were visited every month by a village health worker. Therefore, the provision of a village health worker to each village development committee and increases in the frequency of their visits to communities should lead to large increases in service utilization. By contrast, the influence of community health volunteers on service use is minimal. The results indicate that even the actively working volunteers have modest effects on service uptake. In view of these results, health planners should radically appraise the volunteer programme. Improvements in effectiveness may require the provision of further training, closer supervision and financial incentives. But the costs of these new features should be carefully assessed before their widespread imple-mentation. Endnotes 1 In 1995 (after the completion of the study) regular immunization services were supplemented by a national vertical campaign with designated National Vaccination Days. 2 To avoid statistical complications, one child under five (or pregnancy) was selected at random per mother. 3 Public sector health staff are permitted to hold private clinics outside official working hours. On the assumption that health staff with reasonable skills and a good reputation were more likely to hold private clinics than other health staff, this factor was taken as a possible quality indicator. 4 The typical family was defined as follows: low socioeconomic status, both husband and wife illiterate and two or more births in the past 5 years. 5 The relatively weak effect of access is consistent with the results for the larger study of 237 clusters (not shown here). In this larger study, access to the nearest health post was measured in distance rather than travel time. A threshold effect was found. Uptake of services was lower when the nearest health post was 20 km or more away, but below this distance, no further improvement in utilization occurred as access improved. Only 13% of the 237 clusters were disadvantaged by being 20 km or more away from the nearest health post. References Acharya LB Utilization of family planning and MCH services in rural Nepal: the effects of service access and quality. Unpublished PhD thesis, University of London, UK. Aitkin M, Anderson D, Francis B, Hinde J Statistical modelling in China. Oxford: Oxford University Press. Aitkin M, Longford N Statistical modelling issues in school effectiveness studies. Journal of Royal Statistical Society, Series A 149: Baker J, Freedman R, Thapa S, Rai T Understanding quality of service in family planning in Nepal. Journal of the Nepal Medical Association 32(111): Freedman R Quality of care, infection prevention and other aspects in public sector family planning facilities in Nepal: an action-oriented assessment. Journal of the Nepal Medical Association 34( ): NIV (New Era, IIDs and VARG) Nepal Family Planning, Fertility and Health Survey (NFFHS), Survey Report. Kathmandu, Nepal. UNICEF/HMG Nepal Children and women of Nepal: a situation analysis Kathmandu, Nepal. Woodhouse B, Rasbash J, Goldstein H, Yang M A guide to MLN for new users. Institute of Education, University of London. Biographies Laxmi Bilas Acharya, PhD, is a lecturer at the Central Department of Population Studies, Tribhuvan University, Kathmandu, Nepal. He is also a Population and Development Consultant for the Ministry of Population and Environment and UNFPA Nepal. After completing his Masters in Statistics in 1977, Dr Acharya started his career as an assistant lecturer in the Central Department of Statistics. He joined the Central Department of Population Studies in 1992 after completing a Masters in Demography at the Cairo Demographic Centre. In 1997 he completed his doctoral research on Utilization of Family Planning and MCH services in Rural Nepal: the Effects of Service Access and Quality at the Centre for Population Studies, London School of Hygiene and Tropical Medicine. His research interests include fertility, mortality, reproductive health and health services utilization. John Cleland is Professor of Medical Demography at the London School of Hygiene and Tropical Medicine. Previously, he has worked in the Survey Research Centre at the London School of Economics, the Population Bureau of the UK Overseas Development Administration, in Fiji (on an evaluation of the family planning programme and related research), the World Fertility Survey in London, and the International Statistical Institute in The Netherlands, before joining the School in His career has been concerned almost entirely with the demography of developing countries. His research interests include fertility, family planning, child mortality and risk behaviour in relation to STD/HIV infection. An experienced teacher, he takes responsibility for courses on demographic data collection and family planning programmes which form part of the School s Masters programmes. Correspondence: Prof. John Cleland, Centre for Population Studies, London School of Hygiene and Tropical Medicine, Bedford Square, London, WC1B 3DP, UK.