Whose priorities count? Comparison of community‐identified health problems and Burden‐of‐Disease‐assessed health priorities in a district in Uganda

Abstract

Objectives The aim of the study was to compare health problems as defined quantitatively by the Burden of Disease study to those defined by the community. The secondary aim was to explore the potential for using qualitative participatory methodologies as tools for planing and priority setting.

Design Interviews and group discussions with a purposely sampled set of community members (n=51) and community leaders (n=6). The Nominal group technique, as well as in‐depth interviews, were used to identify major health problems – as perceived by the community. Epidemiological data on the major health problems were derived from the national Burden of Disease study.

Results Community perceived health problems were similar to those identified by the burden of disease study. Reasons given for the ranking included prevalence, fatality, social and cultural stigma. Social stigma and cultural values were not considered in the burden of disease studies. However, socially stigmatized diseases were considered to be more serious compared to non‐stigmatised conditions, in spite of their low prevalence. Poverty and lack of knowledge were the perceived major causes of ill‐health in the community.

Conclusions Qualitative approaches like the nominal group technique may be useful in eliciting community values that could supplement quantitative information like that elicited by the Burden of Disease study. Such a mixed approach would capture both epidemiologicaly assessed and community felt needs in the priority setting process.

Introduction

The Global Burden of Disease study and its associated cost‐effectiveness approach has been used by international organizations and health planners to assess health systems, and to identify priorities for health sector reform. The approach is seen as attractive because it provides a common epidemiological unit for quantifying major health problems. ¹ Disability Adjusted Life Years (DALYs), is the summary measure used to estimate Burden of Disease and it considers both time lost due to premature mortality and time lived with disability. The Burden of Disease study estimates the major causes of DALYs lost and expresses the results in single ranking lists that policy makers can use to identify the country's or districts' major health problems. ² Since DALYs can also be used to assess the cost‐effectiveness of various interventions, the methodology provides a potential tool for planning and resource allocation. ³

Some countries, Uganda included, have attempted to use this approach, carrying out a national and a 13‐districts Burden of Disease study (Table 1). The information on major health problems thus identified was used in the formulation of their national health policies. ⁴ , 5, ^–6

Table 1.

 Uganda National and Mukono District top causes of Burden of Disease

The Ugandan Burden of Disease study was performed in 1994, at the same time as Uganda was decentralizing its health‐care. ⁷ Public participation is one of the stated goals of decentralization.

The public should be involved in identifying their needs if they are expected to support the proposed programmes. Meaningful participation, given the heterogeneity of communities, requires representation of the different groups that exist within communities. ^{8 , 9} The Burden of Disease approach has, however, been thought to be less participatory, involving mainly experts, who may not necessarily represent the values of the public. ¹⁰ It may not have catered for the psychosocial and cultural perceptions of health or the context of disabilities. ¹¹ Participatory, qualitative methods like the nominal group technique, however, provide greater understanding and a direct link with the target population. ¹²

One important motivation for this study was the idea that participation, if incorporated into the Burden of Disease approach, would make the whole enterprise more transparent and could increase public accountability and create legitimacy for the hard choices that have to be made. ^{8 , 9} To our knowledge, there is no in‐depth description or analysis that compares community based ranking of health problems to those obtained by quantitative standards such as the DALY methodology.

The aim of the study was to compare health problems as defined quantitatively by the Burden of Disease study to those defined by the community. The secondary aim was to explore the potential for using qualitative participatory methodologies as tools for planning and priority setting.

Design, materials and methods

We chose a qualitative, exploratory approach, using the nominal group technique and in‐depth interviews. The nominal group technique involves working with small groups to assess community perceptions of problems in a way that overcomes the usual unequal representation of opinions. By compensating for the dynamics of social power, it is more effective in generating ideas and getting equal participation, as compared to other group discussion methods. ⁸

Participants were residents of Nama subcounty in Mukono district. The district is fully decentralized and participated in the national Burden of Disease study. ^{13 , 14}

Five group discussions were convened (a total of 51 people) (Table 2). Each group discussion took 1½–2 h. Participants were mobilized by the community leaders and should have lived in the community for at least 2 years prior to the study. Participants identified a convenient location for the meetings.

Table 2.

 Characteristics of group discussants

Using the nominal group technique, each group discussion involved self‐introductions, listing of participants' expectations, establishment of group norms and introduction of the purpose of the meeting. This was followed with a brain‐storming discussion on introductory questions, such as the determinants of ill‐health and 10–15 min to quietly think about the issues introduced by the facilitator. Using the Round Robin technique, each person was asked to tell the group what they thought were the priority problems in their community (one item at a time). These were listed on a flip chart, in a language understood by the participants. The process was repeated, until all the items each person had, had been listed, but without discussion. Next, each participant gave what they thought should be the first priority and their reasons. Discussions were then held, and the reasons for ranking were recorded. Finally, voting was used to develop a priority list for the top 10 health problems. No cues were given during the discussions. The whole process was written and audio‐taped (with permission from the participants). These results were supplemented with in‐depth interviews.

In‐depth interviews involved Local Council chairmen (LC) and other key informants, selected because of their experience in priority setting. Snowball sampling was employed to identify the six informants. ¹⁵ The interviews were carried out using a piloted interview guide, lasted a minimum of 45 min, and both notes and audio‐recording were used. During the data collection period, the investigator listened to the audio‐taped discussions, and made verifications with participants (where possible).

Analysis

Data analysis involved transcription and translation of audio taped information, with care to preserve the original meaning. The transcribed material was compared to the written information for validation. Interpretations, using the template organizing style, were then carried out. ¹⁶ The templates used to develop codes, which facilitated the break down and re‐organization of data to form categories and themes, were obtained from literature. These form the basis of the report.

Results

The nominal group technique and the interviews enabled us to identify the top 10 health problems, as perceived by the community. The rank order was comparable to the results from the Burden of Disease for this particular district (1, 3Tables 1 and 3).

Table 3.

 Comparison of priority disease conditions* as mentioned by the different groups

Diseases mentioned by at least four groups included dental problems, diarrhoea, measles, malaria, and respiratory infections. A condition was ranked highly if it either affected or killed many people, was costly or difficult to treat at the local health unit. The women considered conditions that affected children. Social consequences of a condition, ease of treatment and prevention were additional considerations. Below we report in more detail the reasons given for the ranking given by the various informants.

AIDS is a big problem; it has killed many

AIDS was thought to be a big problem, rating first in three of the five groups, because of its prevalence, fatality and social consequences (Table 3).

The following remarks were made:

`…AIDS is a big problem, especially among us, the youths… even the way you get it… it is difficult to avoid…' (Youth group)

`…Malaria is really a problem, the suffering it causes to the children is very big…' (Women's group)

`…Yes, malaria affects many people and kills but if it kills the man, the wife can remain and look after the children, but with AIDS, both of you go and you leave the children on their own… AIDS just clears homes….' (Men's group)

`…Infertility may not be common but it brings a lot of misery especially for the women…' (Women's group)

`…Although eye infections and teeth problems are common, they are easy to prevent, you just need to wash your face with salty water and brush your teeth daily…' (Youth group)

Results from the in‐depth interviews were somewhat different (Table 4). The community leaders ranked malaria and respiratory infections highest. Only three respondents mentioned AIDS. Prevalence was their main reason given for the ranking.

Table 4.

 Key informants' disease priorities

Other conditions not mentioned by the respondents

The facilitator introduced another group of conditions not mentioned initially by the respondents. These included conditions from four of the seven disability disease categories used in the Burden of Disease ¹⁷ but are also stigmatized in Uganda (Box 1). Conditions included epilepsy, infertility, psychosis, leprosy and impotence.

Table 5.

Box 1 Ranking of the facilitator introduced conditions (and reasons) as compared to Burden of Disease relative ranking

Although rare, the discussants considered these problems to be worse than all previously discussed conditions (Box 1). Erectile dysfunction was ranked first, followed by infertility, epilepsy and psychosis, respectively. Social and cultural stigma attached to these conditions determined the ranking.

Perceived determinants of ill‐health

During the group sessions, perceived causes of disease and ill‐health in the community were also discussed.

It all stems from poverty and lack of knowledge….

The main perceived causes of ill‐health in the community were poverty and lack of knowledge (Table 5). Additional causes included lack of food and sources of income. The female discussants, however, mentioned specific causes, namely sexual harassment and domestic violence.

Table 5.

 Determinants of ill‐health

The comments included:

`…These men really beat us, they have turned us into drums…' (Women's group)

`…Lack of knowledge is a big problem, even having money may fail to improve your health if you lack knowledge…' (All)

`…poverty is a big problem, without money you cannot do anything, everything requires money, the food, house…' (All)

Discussion

The methodology helped us to tap into people's experiences with health and disease in their community. The nominal group technique was used because it overcomes the usual unequal representation of opinion and being face‐to‐face, allows for explanation and clarification of issues. ¹⁸

However, the approach is limited to community perceptions. Perception can be influenced by media, personal knowledge and experiences and may not necessarily present the true picture. Further, lack of cues may make the results less comparable across groups but allows for a wide range of opinion and minimises bias. The approach is also liable to researcher bias, being dependant on his/her insight and interpretation. Feedback given to the participants during the data collection minimizes this. Hence, it would be futile to base research for health planning only on community perceived needs, but they have the potential for enriching the quantitative data. Additionally, they can be used to explain quantitative results, generate new ideas and aid in developing hypothesis that can be tested quantitatively. ¹²

Disease ranking

The top 10 identified diseases were similar to those identified by the Burden of Disease study, with some variations in actual ranking and reasons for ranking. The differences in ranking of diseases could have been influenced by participants' knowledge and experiences with the diseases mentioned.

This may account for the fact that AIDS was ranked highly while Tuberculosis, also a prevalent problem, was not mentioned at all. On the other hand, if the methodology for collecting quantitative data is not rigorous enough, it also introduces bias, ¹⁹ for example, if only facility based data is used. This calls for triangulation of data sources (facility and community) and methodologies (quantitative and qualitative, participatory) in priority setting.

Some of the reasons given for the ranking of diseases were similar to those considered in Burden of Disease. ² However, other criteria, such as social and cultural stigma, were introduced. This was not catered for in the Burden of Disease study, although we found it to be important in this community, and others report similar results from other African cultures. ²⁰ Further, the ranking of these conditions was opposite to the disability weight used in the Burden of Disease study. Whereas Burden of Disease gave more weight to mental problems compared to sexuality‐related problems, the community ranked sexuality related problems higher. ¹⁷

Determinants of disease and ill‐health

The participants' demonstrated an understanding of the determinants of ill‐health in their community, namely poverty and lack of knowledge. The interaction between health, poverty and lack of education (knowledge) have been well documented. ²¹ However, since some of the mentioned concerns are outside the health sector, there is need to collaborate with other sectors, on these key issues.

An approach that visualizes health as a means to the end, for example the Uganda Health policy, where it is a strategy for poverty alleviation, would legitimize the collaboration. Group‐specific determinants of ill‐health could be partly accounted for by gender. The adolescent girls ranked sexuality problems high, which is consistent with current knowledge, in that by 19 years over 70% of the women have begun child bearing. ²² Violence was an issue for the women and was also identified as an important risk factor in the Burden of Disease study at national level but not at district level. Domestic violence and sexuality problems are some of the known, yet concealed, problems that are unlikely to be identified through quantitative approaches. ²³ These may require methods, for example group discussions, that build trust and freedom to speak. ¹²

Conclusions and recommendations

The methodology employed in this study facilitated eliciting of values and health problems that may have not otherwise been identified. Within the decentralization framework, participation of different groups in the community should be effected using simple and cheap methods like the one employed in this study to supplement the more quantitative approaches. The approach provided a deeper understanding of how the participants thought and felt about various health problems, captured the heterogeneity in the community through wide participation, captured the social and cultural values and impact of disease in this community and could be used to increase public accountability and create legitimacy for hard policy choices.

These values need to be reflected in priority setting. The approach employed in this study can facilitate placing of epidemiological findings in the social and cultural context, thus designing of appropriate and acceptable interventions for the community.

The highlighted determinants of health call for the participation not only the public but also other sectors. A multi‐sectoral approach to health would come a long way in addressing some the key problems including poverty, education and gender inequalities.

The marked overlap of disease priorities is a good foundation for participatory planning of interventions. Cost‐effective interventions for epidemiologically identified priority diseases could be introduced and discussed with the community. However, the subtle differences between groups need to be recognized in planning. For consensus building in priority setting for health, different groups in the community should be involved, using participatory approaches such as the nominal group technique and other consensus building approaches.