Making a case for diversion of solid medical waste from households A generation study in Ga South municipal assembly, Accra, Ghana

Background: Solid medical waste (SMW) is generated from the healthcare industry but can also be found in households when activity involving patient care occurs. Its hazardous properties require special treatment to minimize hazards to the environment. To achieve this, SMW must be safely diverted from households using a systemic approach, which should be informed by the quantities generated and factors associated with generation. Objective: To characterize household SMW in terms of quantity and composition and to describe the factors associated with its generation. Methods: Manual sorting of household waste was conducted in 60 households to measure quantities of SMW and its components in Ga South Municipal Assembly, Accra, Ghana. Sample collection took place in the wet season (October, 2014) and dry season (December, 2014/January 2015). Rates of generation and percentage composition computed. Factors influencing generation were evaluated with non-parametric tests and quantile regression analysis. Statistical significance was set at p < 0.05. Results: Per capita generation of SMW was 1.77 × 10 −3 kg/person/day. Pharmaceutical waste and sharps waste comprised 98% and 2% of SMW respectively. Generation rates were significantly higher in the wet season than in the dry season (z = 3.129, p = 0.002). Households where medical complaints were reported generated significantly less SMW at the 5th, 10th, 25th and 50th quantiles (β = −2.711, p


Introduction
The health care industry generates waste containing pathogens, toxic heavy metals and chemicals, termed healthcare waste (HCW) [1]. The solid component of HCW arising from activities of health protection, diagnosis and treatment is called solid medical waste (SMW) [2]. The traditional domain of SMW is the healthcare industry, where waste is generated from a variety of healthcare activities. However, a limited variety of healthcare activities can also occur in households such as being found in home based care [3] [4], shortened hospital stay [5], care and treatment of chronic diseases in aging populations [6], and home management of illnesses such as malaria [7]. Although SMW constitutes up to 0.1% of the mixed municipal solid waste stream, its hazardous character attracts public sensitivity and poses a challenge to health and municipal authorities [8]. Examples of SMW found in households include discarded medicines, blood soaked bandages, hypodermic needles and syringes, lancets and insulin pens.
Sharps in the waste stream such as hypodermic needles can cause physical injury and may lead to transmission of blood borne pathogens if present. Expired and unused medicines discarded with household waste, often end up in landfills, where household waste is mostly landfilled. Active pharmaceutical ingredients from discarded medicines in landfills can be discharged into leachate [9]. Environmental hazards associated with pharmaceuticals in SMW include destruction of bacteria necessary for sewage treatment, adverse effects on aquatic and terrestrial life, and air pollution when medicines are burnt at low temperatures [10] [11]. Antibiotic resistance has been demonstrated in viable organisms present in untreated landfill leachate [12]. In Ghana, lack of a waste segregation system in most residential premises suggests that SMW is mixed with household waste. In some cases, final disposal occurs at illegal dumpsites by informal waste porters, who often work without protection. When household waste (HSW) is deposited in poorly maintained landfill sites and unauthorized open dump sites, unrestricted access by scavengers and young children exposes them to community acquired needles tick injuries (CANSIs). Among the reported consequences of this exposure, hepatitis B infection is the most frequently reported, although HIV appears to be the most feared [13] [14].
In recognition of these health hazards, some safety measures have been applied elsewhere. For instance, collection of sharps waste is available for households in some parts of the United States [15] [16] [17]. In Sweden, take-back programs facilitate the return of unwanted medication through pharmacies and even if they were to be discarded in household waste, only 1% of household waste is land filled [18]

Sample Selection
A household was the sampling unit in consistency with earlier studies [20] [25] [26]. For waste stream analysis, a minimum of 50 sampling units (households) per 500 households has been suggested by Igbinomwanhia (2011), giving a ratio of 1:10 [27]. On this basis, sixty households were selected from a pre-existing sampling frame of 600 households in October, 2014. In the sampling frame, twenty households each were selected by multi-stage sampling from 30 enumeration areas (EAs) in the municipal assembly during an earlier phase of the field work [28]. Households for the waste stream analysis were selected by ballot.
Two households were selected from each of the 30 EAs, making a total of 60 households. Once a household was selected, a member of the household was informed and consent obtained to collect household waste. All households informed accepted to participate in the study.

Household Waste Collection
Identification numbers were assigned to participating households comprising four digits. The first two digits represented their location (numbered 01 to 30) and the second two digits were their serial numbers from the sampling frame (numbered 01 to 20 in each location

Manual Sorting of Household Waste
The unsorted content of each household's bagged waste was weighed before sorting. To obtain the quantity of specific waste components, manual sorting was undertaken by four trained field staff on a table overlaid with a wire mesh on a clean plastic sheet. The specific waste components were pharmaceutical waste, sharps waste and offensive waste ( Table 1).
The waste components were manually sorted and each fraction was weighed.
Measurements were estimated on the basis of wet waste (w/w) in kilograms (kg)

Waste Survey
A waste survey was conducted by four trained data collectors using a purpose designed 32-item questionnaire to obtain respondent and household characteris-

Data Management and Analysis
Descriptive statistics of the sample households were generated. Other variables computed were household daily waste generation rate, per capita daily generation rate, percentage waste weight (%  [29], except that the upper limit for the middle income category in [29] was GHC500.
The data were analyzed using non-parametric tests. The Kruskal-Wallis H test evaluated variation in daily generated quantities of SMW across assigned income groups. Seasonal variation was assumed a priori in household generation of SMW and evaluated with the Wilcoxon signed ranks test. Household characteristics that might influence generation of SMW were determined in two steps.
First, within-variable differences in household SMW generation were tested using the Wilcoxon rank sum test. Each characteristic namely medical complaints, National Health Insurance Scheme (NHIS) membership, education, presence of under-fives, type of house and room occupancy, had "1" assigned to the risk category and '0' to the reference category. A p-value < 0.05 indicated significant variation in household generation of SMW. In the second step, quantile regression was used for multivariable analysis, applying the statistical model in Equation (1) [30].
where Y is the household generation of SMW (kg/household/day), α is a constant term, β i represents the regression coefficient for ith household characteristic, X i , and the residual error term is represented by ε [30]. P-values were generated with Stata version 14.0 (Stata Corp College Station, USA), and based on the hypothesis that the computed regression coefficient equals zero. A p-value less than 0.05 implied that variability in household generation of SMW was unlikely to be due to chance.

Participant Characteristics
Sixty households were recruited for the household waste stream analysis.
Based on capita generation and population size in the 30 locations (n = 24,183), the average daily quantity of SMW generated was 42.80 kg when outlier measurements were included and 32.40 kg, when they were excluded. No significant variation was found in SMW generated across the income groups per household [H(2) = 1.40, p = 0.497] and per capita [H(2) = 3.08, p = 0.214].

Percentage Composition of Waste Sub-Streams in the Household Waste Stream
During the study period, the average percentage composition of SMW in

Seasonal Variation in Generation Rates of Waste Sub-Streams
When all income groups were combined, the per capita daily generation rates were significantly higher in the wet season than in the dry season for pharmaceutical waste, solid medical waste and offensive waste. Sharps waste showed no seasonal variation (z = 1.938, p = 0.053) ( Table 4).

Factors Associated with Quantity of Solid Medical Waste
Preliminary analysis showed that medical complaints, type of house, and room occupancy might influence the distribution of SMW (Table 5).
After multivariable analysis, reported medical complaints emerged as the only significant factor influencing generation of SMW. Households that reported medical complaints generated significantly less SMW than households that did not report medical complaints, controlling for type of house and room occupancy ( Table 6). The greatest difference was observed at the 75th quantile, when outliers were excluded (β = −5.107; p = 0.005) ( Table 7).

Description of SMW Recovered from Household Waste
Among items recovered from pharmaceutical waste, antibiotics, multivitamins,     in dispensing envelopes, blister packs and as loose tablets. Antihypertensive (nifedipine, lisinopril, amlodipine, bendrofluazide) and antidiabetic (glibenclamide, metformin) medicines were also present in blister packs or dispensing envelopes.
Sharps recovered were predominantly used razor blades. A few disposable shaving sticks were also present. Needles (capped and uncapped) and syringes were recovered from a single household bin bag. Offensive waste mostly comprised of were found on two disposal events. No intravenous tubing or cannulae were seen.

Generation of Solid Medical Waste in Households
At the time of writing, no study had described the quantification of SMW in household solid waste in Ghana, therefore our study represents the first generation study conducted in a local community. Households typically generated between 5 and 7 grams of SMW daily. Although these were relatively small amounts, the total daily production of 32.40 kg computed for the study population is substantial. With the outlier households, the daily production of 42.80 kg represents situations where hoarding of medicines may occur in the household.
This could result in periodic or one-off disposal of large quantities of SMW. The small sample of households used in the study suggests that the values are indicative, but its composition mostly of unwanted medicines, especially antibiotics raises concern. This is given the fact that nearly all household waste in Ghana is sent to landfills.
Household generation of SMW showed significant seasonal variation. This confirmed our assumption a priori. Some diseases that exhibit seasonal variation often require the use of medicines, such as malaria and respiratory tract infections.
This would result in the generation of SMW from left over or expired medicines. Waste generated would include medicines and/or their containers, but not packaging. The higher generation of SMW in the wet season may be partly attributed to common acute illnesses which tend to peak with the rains, such as malaria, respiratory tract infections and some diarrhoeal diseases. The medicines recovered during the waste stream analysis included therapeutic categories often prescribed or bought over the counter for these conditions. Therefore, the consumption of medicines may be higher in these seasons and left over medicines and their containers generate SMW. It is also possible that at the onset of the survey which was in the wet season, a few households may have utilized the opportunity to discard stored waste items since waste collection was offered at no cost to households, whereas the services rendered by the waste management companies had to be paid for monthly. However, these outliers were limited to less than 5% of the sampled households.
Cussiol et al. (2006) sampled municipal solid waste (mostly of residential origin) to quantify potentially infectious waste [31]. Therefore we compared our results with this study. Unwanted medicines referred to as 'chemical waste', accounted for 1.91% of the waste sample in the reference study. We found a lower proportion of 1.05%. It is likely that SMW from other sources may partly account for differences observed in waste composition between the reference study

Potential Hazards to Health and Environment
Unwanted medicines comprised the bulk of SMW in households. The therapeutic categories of the medicines recovered were consistent with acute and chronic diseases prevalent in Ghana. Acute conditions such as malaria, respiratory infections and diarrhoea [32] are prevalent, as well as chronic conditions, such as The greatest concern is the risk of antibiotic resistance as antibiotics were the largest category of medicines recovered from household waste, particularly the penicillin group. In a similar study of municipal solid waste in Florida, USA, antibiotics as a group was found in the largest quantity, followed by non-steroidal anti-inflammatory drugs [37]. Diclofenac and ibruprofen have been reported in earlier studies to be associated with toxic effects in birds [38], in aquatic life [39] and reduces survival of decomposers [40]. Razor blades were the most common type of sharps waste, as also reported in the study by Cussiol et al. (2006) [31].

Limitations of the Study
At the onset of the study, some of the households did not place their household waste for collection as agreed. This changed and collection improved in the later weeks. To compensate for these events, single mean imputation for missing data was used to compute the missing weight measurements in Stata version 14.0 (StataCorp LP, Lakeway Drive, Texas, USA).
Due to the small sample size of households for the waste stream analysis, results can only be considered indicative. To obtain quantitative estimates intended for regional planning, larger samples taken over successive surveys are recommended.
The non-normal distribution of weight measurements of SMW is due to its generation in relatively smaller quantities compared to healthcare facilities. The clustering of measurements close to zero, and fewer extreme values often resulted in a positive skew. The weight of the medicine containers might have affected the weight of SMW, however these were not disregarded as residue left in them can contain active ingredients. Finally, the assignment of income groups arbitrarily, limits the generalization of the results beyond the study location.

Conclusion and Recommendations
Generation of SMW is influenced by medical complaints and is higher in the wet season than in the dry season. As SMW comprised largely of pharmaceuticals, segregation at source could divert this sub-stream for appropriate treatment and disposal to minimize any potential environmental and/or health impact. The relatively smaller quantities of sharps confer some hazardous properties on household waste and should be safely diverted from the waste stream. The impact of continual deposits of SMW generated at computed rates in the study area is unknown, but extant literature and waste composition rationalize concerns about antibiotic resistance and toxicity to wildlife. Therefore, it is pertinent that future policy on the management of SMW takes into account quantities gener-

Declarations Ethical Considerations and Clearance to Conduct the Study
Ethical approval for the study was obtained from the Noguchi Memorial Institute of Medical Research (NMIMR) Institutional Research Board. Written permission to conduct the study was obtained from the Municipal Chief Executive (MCE). Clearance was obtained verbally from community chiefs during a meeting convened by GSMA. At the meeting, the purpose of the study and the stages in the study were briefly explained. Individual informed consent was obtained from all persons prior to questionnaire administration. Respondents were interviewed in the privacy of their homes and only eligible households participated in the study. Participants' rights to withdrawal from the study were upheld and all participants were treated with respect.