Self-rated physical health, health-risk behaviors, and disparities: A cross-sectional study of youth in the slums of Kampala, Uganda

ABSTRACT

Self-rated physical health (SRPH) has been extensively used to assess health status. In this study, we examine how youth living in the slums of Kampala perceive their physical health and the psychosocial correlates of poor health. Cross-sectional data from the 2014 Kampala Youth Survey (N = 1,134) of youth ages 12–18 years was used to conduct the analyses. Chi-square tests and logistic regression analyses were conducted to determine associations between SRPH, demographic and psychosocial characteristics. Overall, 72% of youth rated their health as ‘excellent’ or ‘good.’ Poor SRPH was associated with older age and lower education, but not with sex. Also, orphans (OR = 2.03; 95%CI:1.51–2.72), those who lived on the streets (OR=3.09; 95%CI:2.30–4.15), who did not have electricity (OR = 2.83;95%CI:2.12–3.78), who initiated alcohol use early (OR = 2.08; 95%CI:1.47–2.94), who frequently get drunk (OR = 5.67; 95%CI:2.69–11.96), who were HIV positive (OR = 2.18; 95%CI:1.47–3.23), who had been injured due to their drinking (OR = 2.09; 95%CI:1.44–3.03), who thought about hurting themselves (OR = 2.09; 95%CI:1.60–2.73), and those who often felt lonely (OR = 2.54; 95%CI:1.61–4.02) had higher odds of poor SRPH compared to their peers without these characteristics. Poor SRPH may serve as a marker for multiple health-risk behaviors and severe health disparities among youth in vulnerable and resource-limited settings.

KEYWORDS:

• Youth
• self-rated health
• health disparities
• Uganda
• Africa
• urban

Introduction

Self-rated physical health (SRPH) is a helpful, albeit subjective, measure that has been utilised widely to obtain a person’s self-assessment of their health (Jylhä, 2009). In order to answer the question, an individual must consider their perception and understanding of ‘health’ and how it applies to them within their social, cultural, physical, mental, and biological context (Falk et al., 2017; Meireles et al., 2015; Subramanian et al., 2009). Originally, it was developed to replace a clinical assessment of health for participants in survey research (Strawbridge & Wallhagen, 1999). In the early assessment of the measure and its association with mortality, the authors concluded that it was a ‘deceptively simple variable that likely measures a great deal more than disease burden’ (Strawbridge & Wallhagen, 1999). As a result, SRPH has been used in research across the world both in clinic settings and population surveys (Bombak, 2013), including the World Health Survey by the World Health Organisation (Subramanian et al., 2009a).

Although the measure of SRPH is subjective, it has strong associations with mortality risk and other health outcomes among adults, across settings (Falk et al., 2017; Hirve et al., 2012). For example, in Brazil, the mortality rate was higher among those who rated their SRPH as poor, compared to those with better SRPH (Inuzuka et al., 2018). Similarly, in China, poor SRPH was positively associated with mortality (Shen et al., 2014) and several other health outcomes such as cardio-cerebral diseases and mental illness compared to objective health measures (Wu et al., 2013). Researchers who have studied the validity of SRPH find that it is a useful marker of overall health (Wu et al., 2013) and that it can be used to assess basic gender differences, socioeconomic disparities in population groups (Kuhn et al., 2006) and geosocial pockets of health inequity (Cislaghi & Cislaghi, 2019).

Studies have also used SRPH to understand health perceptions and risk behaviours among youth. In a global study of 32 countries, females consistently rated their health as more ‘poor’ than males, and the disparity increased with age (Cavallo et al., 2015). Also, Rathmann and colleagues examined how academic wellbeing and school climate affected middle school students’ health perceptions and found strong associations between better learning environments and good SRPH (Rathmann et al., 2018). A Swedish study found that the social status of youth was positively associated with SRPH (Joffer et al., 2019). A Brazilian study found that poor SRPH was associated with alcohol consumption and drug experimentation (Malta et al., 2018). In Thailand, poor SRPH was associated with overweight, loneliness, shyness, hopelessness, and low self-rated happiness and seemed to be affected by the financial situation of the family, school achievement, and reports of tobacco usage (Page & Suwanteerangkul, 2009). Finally, a Norwegian study found that self-rated health in adolescence was associated with multiple illnesses in adulthood (Hetlevik et al., 2020).

These previous studies show that SRPH has been used across countries and in different contexts to predict health outcomes, including death, but also to determine links to social status, learning, and other health-factors (Balázs et al., 2018). However, limited research exists of SRPH in sub-Saharan Africa despite researchers urging for its use as a simple tool to assess health status in resource-poor settings (Kuhn et al., 2006). Among youth in low-income countries such as Uganda, the population growth is rapid; nearly 50% of the population is 15 years and younger (Johns Hopkins University, n.d.). Youth living in the slums of Kampala face particular hardships; many are orphans (Swahn et al., 2017), and many are exposed to a range of harm and health risks pertaining to child maltreatment (Swahn et al., 2017), risky sexual practices (Kumar et al., 2020), sexual exploitation and sex work (Self-Brown et al., 2018; Swahn et al., 2016), HIV (Swahn et al., 2019), alcohol use (Swahn et al., 2014a; Swahn et al., 2020), violence (Culbreth et al., 2019; Swahn et al., 2015), and mental health concerns including suicidal ideation (Culbreth et al., 2018).

Within this context, simple and low-cost screening questions and tools that can quickly assess health needs among youth are needed. As such, a measure of SRPH was added to a study on health-risk behaviours among youth in the slums of Kampala to examine the extent to which it may be used to identify disparities and urgent health needs in this vulnerable population group. In this study, we examine how youth living in the slums of Kampala perceive their physical health and the psychosocial correlates associated with poor SRPH. The goal of the study is to determine if SRPH may be used as a marker for risky behaviours and health disparities in vulnerable youth and to identify those in need of more specific screening and support.

Methodology

This study analyzed the Kampala Youth Survey (N = 1134) conducted in 2014 in the slums of Kampala, Uganda, designed to assess risk behaviours, alcohol use, and HIV/STIs among urban youth. Details of the survey have been described previously (Culbreth et al., 2019; Swahn et al., 2018; Swahn et al., 2019). Briefly, the survey was administered via face-to-face interviews and electronic tablets by trained peer educators and social workers to youth who participated in one of six urban Uganda Youth Development Link (UYDEL) drop-in centres. UYDEL, at that time, served thousands of youth daily who lived in slums and the rural areas of Central Uganda. Their ongoing programmes provide psychosocial counseling, peer support, vocational training, and skill-building for youth in need (https://www.uydel.org). The youth participants for this study lived in the urban slums or on the streets of Kampala and ranged in age from 12 to 18 years old. A total of 1,628 youth were approached and informed about the survey, but only 1,497 (92 percent) consented. Due to technical issues involving offline servers, only 1,134 (69.7 percent) surveys were retained for analysis. As such in our analyses and publications of these data, we consider the full sample size as N = 1,134. Ethical approval for this study was provided by the International Review Board at Georgia State University in Atlanta, Georgia, and by the Uganda National Council for Science and Technology (Swahn et al., 2019). No identifying information was collected as part of the study.

Measures

In this study, ‘self-rated physical health,’ was worded as follows ‘How would you rate your physical health?’ Response options were excellent, good, fair, and poor and dichotomized into two categories for analysis: excellent/good and fair/poor.

A range of measures considered to be potential psychosocial and behavioural correlates were included to assess demographic characteristics, housing, alcohol use, HIV/STI, injury/abuse, and mental health. Demographic characteristics included the descriptive variables of gender, age, education, religion, and parental status. The wording of the variables, sample characteristics, and sources of validation are presented in Table 1. More specifically, we used demographic measures and questions regarding living conditions/housing, alcohol use, HIV/STI, injury and violence, and mental health from previous research conducted in a similar population in Kampala, Uganda (Swahn et al., 2012; Swahn et al., 2013; Swahn et al., 2014a, 2014b; Swahn et al., 2015; Swahn et al., 2017). Additional measures regarding education (Satz Questionnaire; Mũkoma et al., 2009); religion (Tumwesigye et al., 2013); housing conditions (Demographic Health Survey, 2011); alcohol use, (AUDIT, Conigrave et al., 1995; Global Student Health Survey, 2016; Selzer et al., 1975); mental health (Dahlberg et al., 2005). Some of the original questions were slightly altered and response options truncated, for participants to answer more easily.

Table 1. Demographic and Psychosocial Characteristics, Proportion of Participants, and Wording of Measures Used in the Kampala Youth Survey.

Variables (N = 1,132)			Total	%
	Self-Rated Physical Health	Excellent/Good	815	72.0
		Fair/Poor	317	28.0
		Total	1132	100.0
Demographics	Sex	Male	497	43.9
		Female	634	56.0
		Total	1131	99.9
	Age	12–14years old	236	20.8
		15–18 years old	896	79.0
		Total	1132	100.0
	Education	Never Completed Primary	396	35.0
		Completed Primary	263	23.2
		Secondary or Higher	459	40.5
		Total	1118	98.8
	Religion	Christian/Muslim	1057	93.4
		African Trad/Other	75	6.6
		Total	1132	100.0
	Parents	At least 1 parent alive	881	77.8
		Both parents are dead	251	22.2
		Total	1132	100.0
Housing	How many rooms does your household use for sleeping?	One	469	41.4
		Two	365	32.2
		More than two	265	23.4
		Other	32	2.8
		Total	1131	99.9
	What is the roof of your home made of?	Thatch/straw	24	2.1
		Iron sheets	1082	95.6
		Tiles	10	0.9
		Other	16	1.4
		Total	1132	100.0
	Have you ever lived on the streets with no other place to go?	Yes	249	22.0
		No	882	77.9
		Total	1131	99.9
	Do you have any of the following in your home? Electricity	Yes	867	76.6
		No	262	23.1
		Total	1129	99.7
Alcohol Consumption	How old were you when you had your first full drink of alcohol? (more than a few sips)	1–14 yo	174	15.4
		15–18 yo	231	20.4
		Never	727	64.2
		Total	1132	100.0
	How many full drinks containing alcohol do you have in a typical day when you are drinking?	No drinks	787	69.5
		1–2 drinks	195	17.2
		3 or more drinks	150	13.3
		Total	1132	100.0
	Have you ever sought help for your drinking?	Yes	59	5.2
		No	1073	94.8
		Total	1132	100.0
	In the past month: How many days did you drink so much that you were really drunk?	0 days	854	75.4
		1 or 2 days	168	14.8
		3–5 days	78	6.9
		6+ days	32	2.8
		Total	1132	100.0
HIV/STI	Have you been told by a doctor/nurse or HIV counselor that you have a sexually transmitted infection such as syphilis, herpes, bola, or gonorrhea?	Yes	419	37.0
		No	710	62.7
		Total	1129	10.2
	Have you been told by doctor/nurse or HIV counselor that you have HIV?	Yes	116	10.2
		No	986	87.1
		Total	1102	97.3
Injury/Violence	Have you been seriously injured or hurt due to your drinking?	Yes	132	11.7
		No	1000	88.3
		Total	1132	100.0
	In the past year, did your boyfriend/girlfriend hit, slap, or hurt you?	Yes	156	13.8
		No	976	86.2
		Total	1132	100.0
Mental Health	In the past year, did you ever feel so sad or hopeless almost every day for two weeks or more in a row that you stopped doing your usual activities?	Yes	652	57.6
		No	475	42.0
		Total	1127	99.6
	In the past year, did you ever think of hurting yourself?	Yes	394	34.8
		No	731	64.6
		Total	1125	99.4
	In the past month, how often … Have you felt lonely?	Never	214	18.9
		Sometimes	740	65.4
		Often	173	15.3
		Total	1127	99.6
	In the past month, how often … Have you been so worried about something that you wanted to use drugs or alcohol to feel better?	Never	710	62.7
		Sometimes	257	22.7
		Often	159	14.0
		Total	1126	99.5
	In the past month, how often … Did you feel hopeful about the future?	Never	135	11.9
		Sometimes	484	42.8
		Often	508	44.9
		Total	1127	99.6
	Overall, what do you think about the following statement?
I will probably die before I am thirty.	Yes	146	12.9
		No	983	86.8
		Total	1129	99.7
	Overall, what do you think about the following statement?
There are people I can count on in this neighbourhood.	Yes	914	80.7
		No	214	18.9
		Total	1128	99.6
	Overall, what do you think about the following statement?
I feel safe in this neighbourhood.	Yes	875	77.3
		No	253	22.3
		Total	1128	99.6

Statistical analyses

Chi-square and logistic regression analyses were computed using the SAS 9.4 software. All p-values were assessed using a significance level of 0.05. Odds ratios and corresponding confidence intervals for bivariate and multivariate analysis were calculated using logistic regression. Odds ratios with 95 percent confidence intervals containing the value of ‘1’ between the upper and lower bounds were considered not statistically significant. Of the 1,134 youth participants, two were excluded due to missing responses on SRPH.

Two logistic regression models were created to analyze the data (Table 3). The first bivariate model (crude OR) included analysis between SRPH and each demographic and psychosocial variable. The second multivariable model (aOR₁) examined associations between SRPH and each variable and controlled for demographic variables (age, sex, education, and religion).

Results

Among participants, 72 percent responded that their health was excellent or good and 28 percent that their health was fair or poor. The sample distribution and characteristics are presented in Table 1. The sample comprised slightly more than half females (56%) and the majority of participants were 15–18 years of age (79%). Overall, 22 percent of participants were orphans and 40.5 percent had completed secondary school or higher. Also, 22 percent had lived on the streets, 10.2 percent disclosed a positive HIV diagnosis, 13.8 percent indicated dating violence, and 34.8 percent indicated that they had suicidal ideation.

The demographic characteristics, psychosocial risk factors, and health risk behaviours that varied by self-rated health are presented in Table 2. Based on the chi-square tests between each variable and SRPH, only two of the twenty-five factors examined were not statistically significant (gender and religion). For example, orphans, those living in one room, those who did not have electricity, those who had lived on the streets, those who initiated alcohol early, those who typically drink larger quantities of alcohol, those with a positive HIV diagnosis, those who reported dating violence, those who were sad or who thought of injuring themselves, and those who did not think they would live past age 30, reported a higher percentage of poor SRPH than their peers.

Table 2. Comparative Analysis of Psychosocial and Behavioural Characteristics and Self-Rated Physical Health.

	Self-Rated Physical Health
	EXCELLENT/GOOD N = 815 (72%)	FAIR/POOR N = 317 (28%)	Total (%)	P
Sex
Male	366 (44.96)	131 (41.32)	497 (43.94)	0.2682
Female	448 (55.04)	186 (58.68)	634 (56.06)
Age
Mean (IQR)	16.27 (3)	15.83 (4)	1132	0.0008^1
Education
Never Completed Primary	254 (31.44)	142 (45.81)	396 (35.42)	<.0001
Completed Primary	194 (24.01)	69 (22.26)	263 (23.52)
Secondary or Higher	360 (44.55)	99 (31.94)	459 (41.06)
Religion
Christian/Muslim	763 (93.62)	294 (92.74)	1,057 (93.37)	0.5950
African Trad/Other	52 (6.38)	23 (7.26)	75 (6.63)
Parents
At least 1 parent alive	664 (81.47)	217 (68.45)	881 (77.83)	<.0001
Both parents are dead	151 (18.53)	100 (31.55)	251(22.17)
How many rooms does your household use for sleeping?
One	314 (38.53)	155 (49.05)	469 (41.47)	0.0011
Two	273 (33.50)	92 (29.11)	365 (32.27)
More than two	209 (25.64)	56 (17.72)	265 (23.43)
Other	19 (2.33)	13 (4.11)	32 (2.83)
What is the roof of your home made of?
Thatch/straw	10 (1.23)	14 (4.42)	24 (2.12)	<.00012
Iron sheets	788 (96.69)	294 (92.74)	1,082 (95.58)
Tiles	9 (1.10)	1 (0.32)	10 (0.88)
Other	8 (0.98)	8 (2.52)	16 (1.41)
Have you ever lived on the streets?
Yes	131 (16.09)	118 (37.22)	249 (22.02)	<.0001
No	683 (83.91)	199 (62.78)	882 (77.98)
Do you have any of the following in your home? Electricity
Yes	670 (82.41)	197 (62.34)	867 (76.79)	<.0001
No	143 (17.59)	119 (37.66)	262 (23.21)
How old were you when you had your first full drink of alcohol?
1–14 yo	103 (12.64)	71 (22.40)	174 (15.37)	0.0001
15–18 yo	166 (20.37)	65 (20.50)	231 (20.41)
Never	546 (66.99)	181 (57.10)	727 (64.22)
How many full drinks containing alcohol do you have in a typical day when you are drinking?
No drinks	589 (72.27)	198 (62.46)	787 (69.52)	0.0045
1–2 drinks	125 (15.34)	70 (22.08)	195 (17.23)
3 or more drinks	101 (12.39)	49 (15.46)	150 (13.25)
Have you ever sought help for your drinking?
Yes	34 (4.17)	25 (7.89)	59 (5.21)	0.0116
No	781 (95.83)	292 (92.11)	1,073 (94.79)
How many days did you drink so much that you were really drunk?^4
0 days	639 (78.40)	215 (67.82)	854 (75.44)	<.0001
1 or 2 days	114 (13.99)	54 (17.03)	168 (14.84)
3–5 days	51 (6.26)	27 (8.52)	78 (6.89)
6+ days	11 (1.35)	21 (6.62)	32 (2.83)
Have you been told by a doctor/nurse or HIV counselor that you have a sexually transmitted infection such as syphilis, herpes, bola, or gonorrhea?
Yes	282 (34.69)	137 (43.35)	419 (37.11)	0.0068
No	531 (65.31)	179 (56.65)	710 (62.89)
Have you been told by doctor/nurse or HIV counselor that you have HIV?
Yes	65 (8.23)	51 (16.35)	116 (10.53)	<.0001
No	725 (91.77)	261 (83.65)	986 (89.47)
Have you been seriously injured or hurt due to your drinking?
Yes	76 (9.33)	56 (17.67)	132 (11.66)	<.0001
No	739 (90.67)	261 (82.33)	1,000 (88.34)
Did your boyfriend/girlfriend hit, slap, or hurt you?^3
Yes	101 (12.39)	55 (17.35)	156 (13.78)	0.0298
No	714 (87.61)	262 (82.65)	976 (86.22)
Did you ever feel so sad or hopeless almost every day for two weeks or more in a row that you stopped doing your usual activities?^3
Yes	447 (55.05)	205 (65.08)	652 (57.85)	0.0022
No	365 (44.95)	110 (34.92)	475 (42.15)
Did you ever think of hurting yourself?^3
Yes	245 (30.21)	149 (47.45)	394 (35.02)	<.0001
No	566 (69.79)	165 (52.55)	731 (64.98)
Have you felt lonely?^4
Never	173 (21.33)	41 (12.97)	214 (18.99)	0.0003
Sometimes	530 (65.35)	210 (66.46)	740 (65.66)
Often	108 (13.32)	65 (20.57)	173 (15.35)
Have you been so worried about something that you wanted to use drugs or alcohol to feel better?^4
Never	539 (66.54)	171 (54.11)	710 (63.06)	0.0003
Sometimes	172 (21.23)	85 (26.90)	257 (22.82)
Often	99 (12.22)	60 (18.99)	159 (14.12)
Did you feel hopeful about the future?^4
Never	88 (10.85)	47 (14.87)	135 (11.98)	0.0010
Sometimes	330 (40.69)	154 (48.73)	484 (42.95)
Often	393 (48.46)	115 (36.39)	508 (45.08)
I will probably die before I am thirty.
Yes	88 (10.82)	58 (18.35)	146 (12.93)	0.0007
No	725 (89.18)	258 (81.65)	983 (87.07)
There are people I can count on in this neighbourhood.
Yes	675 (83.13)	239 (75.63)	914 (81.03)	0.0039
No	137 (16.87)	77 (24.37)	214 (18.97)
I feel safe in this neighbourhood.
Yes	643 (79.19)	232 (73.42)	875 (77.57)	0.0370
No	169 (20.81)	84 (26.58)	253 (22.43)

Note. P-values were calculated from chi-square tests unless otherwise noted. P-values were assessed based on a significance level of 0.05. Column percentages used.

¹ Wilcoxon Two-Sample Test;

² Fischer’s Exact Test;

³ Past Year;

⁴ Past Month

Two sets of logistic regression analyses further examined the association between each variable and poor SRPH. In the bivariate model (crude OR; Table 3), age, education, and parental status were associated with poor SRPH. Similarly, variables indicating poverty, alcohol use, HIV status, injury/abuse experience, and mental health characteristics were also generally associated with poor SRPH. More specifically, the strongest associations were observed for orphans (OR = 2.03; 95%CI:1.51–2.72), those who lived on the streets (OR = 3.09; 95%CI:2.30–4.15), those who did not have electricity (OR = 2.83;95%CI:2.12–3.78), and those who initiated alcohol use early (OR = 2.08; 95%CI:1.47–2.94), those who frequently drink until drunk (OR = 5.67; 95%CI:2.69–11.96), those who had a positive HIV diagnosis (OR = 2.18; 95%CI:1.47–3.23), those who had been injured due to their drinking (OR = 2.09; 95%CI:1.44–3.03), those who thought about hurting themselves (OR = 2.09; 95%CI:1.60–2.73), and those who often felt lonely (OR = 2.54; 95%CI:1.61–4.02) as they had higher odds of poor SRPH than their peers without these characteristics.

Table 3. Multivariable Logistic Regression Analyses of the Associations between Psychosocial and Behavioural Characteristics and Poor Self-Rated Physical Health.

	Crude OR (95% CI)	AdjustedOR^1 (95% CI)
Sex
Male	1.00	–
Female	1.16 (0.89, 1.51)	–
Age	0.88 (0.82, 0.94)	–
Education
Never Completed Primary	1.00	–
Completed Primary	0.64 (0.45, 0.90)	–
Secondary or Higher	0.50 (0.36, 0.67)	–
Religion
Christian/Muslim	1.00	–
African Trad/Other	1.15 (0.69, 1.91)	–
Parents
At least 1 parent alive	1.00	1.00
Both parents are dead	2.03 (1.51, 2.72)	1.91 (1.41, 2.59)
How many rooms does your household use for sleeping?
One	1.00	1.00
Two	0.68 (0.50, 0.93)	0.69 (0.51, 0.95)
More than two	0.54 (0.38, 0.77)	0.60 (0.41, 0.86)
Other	1.39 (0.50, 0.93)	1.23 (0.57, 2.63)
What is the roof of your home made of?
Thatch/straw	1.00	1.00
Iron sheets	0.27 (0.12, 0.61)	0.23 (0.09, 0.54)
Tiles	0.08 (0.01, 0.73)	0.06 (0.01, 0.61)
Other	0.71 (0.20, 2.55)	0.55 (0.15, 2.06)
Have you ever lived on the streets with no other place to go?
Yes	3.09 (2.30, 4.15)	3.41 (2.60, 4.67)
No	1.00	1.00
Do you have any of the following in your home? Electricity
Yes	1.00	1.00
No	2.83 (2.12, 3.78)	2.68 (1.99, 3.62)
How old were you when you had your first full drink of alcohol?
1–14 yo	2.08 (1.47, 2.94)	2.41 (1.68, 3.46)
15–18 yo	1.18 (0.85, 1.65)	1.55 (1.07, 2.25)
Never	1.00	1.00
How many full drinks containing alcohol do you have in a typical day when you are drinking?
No drinks	1.00	1.00
1–2 drinks	1.67 (1.19, 2.33)	1.98 (1.39, 2.83)
3 or more drinks	1.44 (0.99, 2.11)	1.83 (1.22, 2.75)
Have you ever sought help for your drinking?
Yes	1.97 (1.15, 3.35)	2.30 (1.32, 4.02)
No	1.00	1.00
How many days did you drink so much that you were really drunk?^2
0 days	1.00	1.00
1 or 2 days	1.41 (0.98, 2.02)	1.69 (1.16, 2.47)
3–5 days	1.57 (0.96, 2.57)	1.96 (1.17, 3.27)
6+ days	5.67 (2.69, 11.96)	6.79 (3.14, 14.69)
Have you been told by a doctor/nurse or HIV counselor that you have a sexually transmitted infection such as syphilis, herpes, bola, or gonorrhea?
Yes	1.44 (1.11, 1.88)	1.49 (1.12, 1.97)
No	1.00	1.00
Have you been told by doctor/nurse or HIV counselor that you have HIV?
Yes	2.18 (1.47, 3.23)	2.25 (1.49, 3.40)
No	1.00	1.00
Have you been seriously injured or hurt due to your drinking?
Yes	2.09 (1.44, 3.03)	2.42 (1.63, 3.58)
No	1.00	1.00
Did your boyfriend/girlfriend hit, slap, or hurt you?^3
Yes	1.48 (1.04, 2.12)	1.71 (1.18, 2.48)
No	1.00	1.00
Did you ever feel so sad or hopeless almost every day for two weeks or more in a row that you stopped doing your usual activities?^3
Yes	1.52 (1.16, 1.99)	1.66 (1.25, 2.20)
No	1.00	1.00
Did you ever think of hurting yourself?^3
Yes	2.09 (1.60, 2.73)	2.26 (1.71, 2.99)
No	1.00	1.00
Have you felt lonely?^2
Never	1.00	1.00
Sometimes	1.67 (1.15, 2.44)	1.70 (1.16, 2.49)
Often	2.54 (1.61, 4.02)	2.51 (1.55, 4.05)
Have you been so worried about something that you wanted to use drugs or alcohol to feel better?^2
Never	1.00	1.00
Sometimes	1.56 (1.14, 2.13)	1.68 (1.22, 2.33)
Often	1.91 (1.33, 2.75)	2.08 (1.41, 3.06)
Did you feel hopeful about the future?^2
Never	1.00	1.00
Sometimes	0.87 (0.58, 1.31)	0.89
Often	0.55 (0.36, 0.83)	0.57 (0.38, 0.88)
I will probably die before I am thirty.
Yes	1.85 (1.29, 2.66)	1.97 (1.36, 2.85)
No	1.00	1.00
There are people I can count on in this neighbourhood
Yes	1.00	1.00
No	1.59 (1.16, 2.18)	1.55 (1.12, 2.14)
I feel safe in this neighbourhood.
Yes	1.00	1.00
No	1.38 (1.02, 1.86)	1.43 (1.05, 1.94)

Note. 95% confidence intervals provided in parenthesis. aOR =  ‘adjusted odds ratio’

Self-rated physical health was coded as 0 = good/excellent health and 1 =  fair/poor health

¹ Adjusted for age, gender, religion, and education

² Past Month

³ Past Year

The multivariable logistic regression model, adjusted for age, sex, religion, and education, examined the associations between psychosocial characteristics, risk behaviours and poor SRPH are also presented in Table 3. The strongest associations were similar to those just noted in the crude, bivariable analyses. However, a few associations varied substantially in the adjusted analyses and need to be highlighted. For example, typically drinking 3 or more drinks in a day was not associated with poor SRPH in the bivariable analyses, but was significant in the adjusted model (aOR = 1.83; 95%CI:1.22–2.75). Similarly, drinking to drunkenness on 1–2 days or 3–5 days were not associated with poor SRPH in bivariable analyses (drinking to drunkenness 6 or more days were statistically significant), but emerged as significant in the adjusted models (aOR = 1.69; 95%CI:1.16–2.47) and (aOR = 1.96; 95%CI:1.17–3.27), respectively.

Discussion

SRPH has been extensively used to assess health status. In this study, we examine how youth living in the slums of Kampala perceive their physical health and the correlates of poor health. Our findings show that 72 percent of our participants rated their health as excellent/good which is substantially lower to many other studies of youth, except for one specific study of vulnerable Roma youth, ages 13, in the Roma settlements in Hungary (Sárváry et al., 2019). In that study, excellent/good SRPH was reported by 65.5% of boys and 58.1% of girls. However, studies in general populations studies in other European countries, such as the Czech Republic for example, find a much higher percentage of youth ages 11–15 years (87.6%) indicating that their health is excellent/good (Hodačová et al., 2017). Additionally, a study of Norwegian youth ages 13–19 found that a high percentage, 89%, indicated that their SRPH was very good/good (Torsheim et al., 2018). These findings overall, indicate that the percentage of youth in our study who perceived their SRPH as good/excellent is relatively low given comparisons with similar age groups in other countries and likely reflects their living conditions, hardships and other forms of adversity.

Previous studies have found important sex differences in SRPH, with females rating their SRPH substantially lower than that of males (Cavallo et al., 2015; Darviri et al., 2011, 2018; Hodačová et al., 2017;Todorova et al., 2013; Vingilis et al., 2002). However, in our study, there were no significant sex differences noted in SRPH and there is no clear explanation for this finding. One possibility, based on previous research that attribute gender differences in SRPH to girls reproductive health issues, and that girls often have higher levels of psychological distress, and greater preoccupation with other health matters (Tremblay et al., 2003), may not be reflected in our study of vulnerable girls. Perhaps in the context of living in dire poverty, the health concerns noted in previous research may be perceived as less relevant or not manifested for the girls, and as such their levels of SRPH is practically the same as the boys. However, this potential explanation is highly speculative and one that we cannot empirically address with the current data. Future research of similarly vulnerable youth should examine gender differences more specifically.

Other key findings from our study confirm previous research by also demonstrating that social and educational inequalities and disadvantage such as poverty, homelessness and orphan status significantly increased the odds of poor self-reported health. In our study, orphans represented 22 percent of our sample, and they were also more likely than their peers to rate their health as poor. Previous research has demonstrated that orphans are more likely to face poverty, less social support and other social inequalities (Page & Suwanteerangkul, 2009; Ssewamala et al., 2009; Swahn et al., 2017). Studies of youth in Scandinavian countries have demonstrated that social inequality, including less family wealth, is associated with higher levels of poor SRPH (Torsheim et al., 2018). Similarly, previous research has also shown that social and other forms of inequality adversely impact SRPH (Darviri et al., 2011; Lau & Ataguba, 2015; Mechanic & Hansell, 1987; Todorova et al., 2013; Vingilis et al., 2002). A recent study also found that reading and writing difficulties specifically, were associated with lower SRPH in Danish adolescents 15–16 years of age, further indicating that specific educational shortcomings exacerbate poor health outcomes (Kjeldsen et al., 2019).

A recent study of U.S. adults also examined material hardship and self-rated health, which may be partly explained by perceived stress but also impact health outcomes differently, particularly in terms of poor SRPH, depression, sleep problems and even suicidal thoughts (Huang et al., 2021). In our study, we included several mental health measures and indicators, most of which were associated with poor SRPH. For example, loneliness, a measure of social isolation was strongly associated with poor SRPH as were feeling hopeless about the future, and thinking of hurting oneself which may indicate a range of unaddressed mental health concerns in our study population. Previous studies have demonstrated high unmet mental needs and psychological distress in this population. (Culbreth et al., 2018; Perry et al., 2020; Swahn et al., 2012). These findings are validated by research in Norway which has reported associations between poor SRPH and social anxiety in youth (Jystad et al., 2021). As such, there is growing empirical evidence that links poor SRPH with mental health indicators across study populations which warrant further examinations.

Surprisingly, and related to the mental health factors just discussed, the strongest correlate of poor SRPH was frequent drunkenness. In this study, having six or more drunk days a month was associated with poor SRPH (aOR = 6.79;95%CI: 3.14–14.69). Intriguingly, several other alcohol measures were also strongly associated with poor SRPH such as early alcohol use initiation, heavy drinking, being injured or hurt due to drinking, and seeking help for drinking. A recent study of Portuguese adolescents found that youth who did not drink alcohol had better SRPH (Marques et al., 2019). Even so, the mechanisms or pathways linking alcohol use to poor SRPH is not well understood and need to be explored, particularly in these young populations with unique vulnerabilities. It is also important to note that alcohol use is also described by youth (Swahn et al., 2014c) as a coping strategy for stigma, psychological distress, suicidal ideation and self-harm (Bousoño et al., 2019; Brick et al., 2018; Rossow et al., 2007; Swahn et al., 2019). However, the strong association observed between poor SRPH and alcohol use needs further research to understand the underlying mechanisms as well as the potential direct or indirect effect in this study. Moreover, given the strong association between heavy and frequent alcohol use and poor SRPH strategies for how to mitigate the health risk will be a key priority given that most of the youth in this study was also below the legal drinking age of 18 years in Uganda.

We also examined a range of other health risk behaviours and their association with poor SRPH. Our findings show that a positive HIV status or reporting other STIs, dating violence and injuries were all associated with poor SRPH. We could not find these factors examined in other youth populations, but suspect that they may exacerbate poor health in other populations and settings as well. These factors, taken together, may indicate a potential constellation of high-risk behaviours that directly and indirectly exacerbate physical health, even in this group of youth ages 12 to18 years.

Several limitations should be considered when interpreting the findings. First, the findings are based on responses from the youth at the UYDEL drop-in centres in Kampala. Due to our large sample size, they may generalise to help-seeking youth in comparable settings and conditions in Sub-Saharan Africa. However, these generalisations may be limited because of our sampling methods and the lack of available studies from similar populations. Moreover, some of the health risk behaviours examined may have been under-reported due to social desirability biases. Second, the SRPH measure is a subjective measure and we did not corroborate any of the health factors examined. Third, while all participants live in the slums, there is substantial heterogeneity in their backgrounds and social status. Finally, this study is based on a cross-sectional survey and causality cannot be determined.

There are several key points to note from the current study. First, despite the hardships and vulnerabilities faced among the youth who live in the slums of Kampala, surprisingly, the majority of youth in our study (72%) consider themselves to be physically healthy, albeit at lower levels, than youth in other countries (Hodačová et al., 2017; Joffer et al., 2019; Meireles et al., 2015; Rathmann et al., 2018; Torsheim et al., 2018; Vingilis et al., 2002). Second, our study shows that poor SRPH is associated with a range of psychosocial risk factors and mental health concerns, particularly frequent and heavy alcohol use, in this vulnerable population of youth. Third, a SRPH measure may be an effective but underutilised tool for assessing health and disparities. Despite being a cross-sectional study, the findings show that this underestimated measure has demonstrated strong associations with behaviours affecting youth and has highlighted areas for additional targeted research and direction for intervention, specifically for heavy and frequent alcohol use. Fourth, developing a short screening tool as a brief measure for overall health status might be prudent in similar low-resource settings. This tool can be utilised to evaluate health programmes and situations where youth may interact with peer educators or settings where they may not be willing to disclose sensitive information. Finally, our findings highlight the range of risk behaviours and health concerns that face these vulnerable youth. Utilising interventions that address these health concerns and co-occurring risk behaviours may be beneficial in low resource settings (Hale et al., 2014).

In terms of future research, our measures on social cohesion need to be examined further. Participants that did not feel social support from their neighbourhood and did not feel safe were also more likely to rate their health as poor. In contrast, youth that felt hopeful about the future were less likely to rate their health as poor. These findings are consistent with prior studies where social cohesion was associated with SRPH among youth (Dageid & Grønlie, 2013; Lau & Ataguba, 2015; Olamijuwon et al., 2018; Stafford et al., 2011). However, given the dire poverty in the slums of Kampala, social cohesion may be considered more modifiable than some of the other risk factors for poor SRPH noted in our study. As such, it may be considered as a potential intervention strategy perhaps coupled with a focus on alcohol use prevention, which in turn may impact multiple risk behaviours and health outcomes. In conclusion, the youth in the slums of Kampala, as a group, face many pressing hardships, including grave poverty and orphanhood. These living conditions are exacerbated by a range of health concerns and risk behaviours that are already impacting their physical health and that will adversely impact their long-term health and longevity, if not addressed.

Word Count: 3,643

Data availability:

For access to the data set associated with this paper, please contact Monica H. Swahn, Ph.D. at [email protected].

Acknowledgments

This research was conducted by the lead author, Annabel Patterson, as part of an MPH Thesis in the School of Public Health at Georgia State University. The research reported in this manuscript was supported by the National Institute on Alcohol Abuse and Alcoholism of the National Institutes of Health under Award Number R21AA22065 to Dr. Swahn. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This work was supported by National Institute of Health: [Grant Number R21AA22065].