Reproducibility,longitudinal validity and interpretability of the Disease Burden Morbidity Assessment in people with chronic disease

Abstract

Objective

To evaluate the reproducibility, longitudinal validity, and interpretability of the disease burden morbidity assessment in people with chronic conditions including multimorbidity.

Methods

The study was conducted using a longitudinal cohort design. A large consecutive sample of adult patients at an Australian community-based rehabilitation service was included with testing at baseline and three-month follow-up (testing longitudinal validity and interpretability). A smaller subsample of patients completed a one-week test–retest (testing reproducibility). Outcome measures included the Disease Burden Morbidity Assessment and 36-item Short-Form Health Survey. Participants in the study received tailored, interdisciplinary intervention between baseline, and three-month follow-up but did not typically receive intervention between baseline and retest.

Results

The longitudinal validity and interpretability sample included 351 participants and the reproducibility sample included 56 participants. Longitudinal validity and interpretability were generally supported with hypotheses supported or partly supported and a small percentage of lowest total scores for impact on daily activities (0.6% at baseline, 1.3% at three-month follow-up). Reproducibility parameters were acceptable for the total score measuring impact on daily activities (e.g. ICC = 0.76).

Discussion

Reproducibility, longitudinal validity, and interpretability of the disease burden morbidity assessment were generally supported for community-based chronic disease patients.

Keywords

Introduction

The patient-reported Disease Burden Morbidity Assessment (DBMA),¹ by Bayliss et al., was designed to measure the disease burden of chronic conditions by rating the impact of chronic conditions on daily activities. This is important as patients have identified daily activities as an essential domain of quality of life,² that may influence a patient’s capacity to participate in healthcare interventions that ultimately reduce the impact of their illness.^3,4 Patient-reported comorbidity questionnaires are appropriate for population³ and community-based research⁵, and are the only method available to obtain information on disease burden over time.⁶ Further, the potential confounding effects of disease burden may need to be adjusted when an index condition is of interest¹ or weighted to account for the severity of the impact of comorbidity.^7–9

Testing of comorbidity measures has predominantly focused on predicting mortality or HRQOL^10,11 or for diagnosis.¹¹ Monitoring the impact of disease burden is also important in healthcare settings as this information may influence the intervention provided and has the potential to reduce further comorbidity. Central to psychometric testing of the DBMA for an evaluative purpose is the conceptualisation of disease burden and multimorbidity (both complex constructs which can be measured using the DBMA). The design of the DBMA is consistent with multimorbidity defined as the coexistance of two or more conditions within a person without reference to an index condition^12,13 as no index condition needs to be listed. Measurement in settings where care focuses on the patient as a whole rather than prioritising one condition are also consistent with this conceptualisation.¹² Of importance in testing for an evaluative purpose is considering prior research that examines temporal patterns of disease burden and multimorbidity. This includes work that indicates congestive heart failure, diabetes, and chronic respiratory conditions predict clinically significant decline in SF-36 physical components scores.¹⁴ As few studies of temporal patterns exist, other work may be useful to inform this testing. For example, comorbidity measures that incorporate disease burden have been found to be stronger predictors of health-related quality of life than measures that incorporate a simple count of comorbidities.^1,3,15–17

Testing of the psychometric properties of the DBMA has primarily been conducted by the original authors using people receiving primary, specialist or hospital care¹ and by independent researchers using a French version of the measure in people attending a general practice clinic.¹⁸ The DBMA has previously been found to correlate more strongly with overall health status, physical functioning, depression and self-efficacy than the Charlson Index, the Risk Adjustment (RxRisk) score or counts of individual conditions,¹ supporting validity of the measure. However, no known study exists comparing the DBMA with other patient-report comorbidity questionnaires or determining the quality of the original measure when used for an evaluative purpose (measuring change over time). Test–retest reliability of the total score of the French version of the measure – the DBMA-Fv has been found to be acceptable (Intraclass Correlation Coefficient (ICC), 0.86; Confidence Interval (CI), 0.79–0.92) in patients aged 18 years or older attending a general practice clinic.¹⁸ The aim of this study was to evaluate the reproducibility, longitudinal validity (responsiveness), and interpretability of the DBMA in adults with chronic disease (predominantly with multimorbidity).

Methods

Study design and participants

A prospective consecutively sampled cohort of people with chronic conditions was used with data collected as part of a larger longitudinal study of people attending a community-based, subacute, interdisciplinary rehabilitation clinic in Australia.⁵ The study setting and characteristics of participants are described in Table 1. Participants with baseline (first attendance at the clinic) and three-month follow-up data from the larger study were used to investigate longitudinal validity and interpretability (Figure 1). Longitudinal validity was defined as correlation between changes in the DBMA and related and non-related constructs (i.e. the ability to detect real change in the construct being measured) rather than treatment effect.¹⁹ To investigate the reproducibility of the DBMA, a smaller consecutively sampled subsample of the study cohort were used. This subsample completed the DBMA at baseline and again one week later with no clinical interventions typically received between these assessments at the interdisciplinary rehabilitation clinic. However, these participants, as well as the remainder of the study cohort typically received tailored, interdisciplinary intervention between the baseline assessment and three-month follow-up. This intervention included individual sessions with allied health professionals as well as programs targeting presenting conditions (e.g. gym-based exercise, back care). A sample size estimate of a minimum of 50 participants was required to detect an ICC of .80 with a 95% confidence interval (CI) from .70 to .90²⁰ and has been reported as an acceptable sample size for testing reproducibility, responsiveness, and construct validity.²¹ Ethical approval was granted by the Central Queensland Human Research Ethics Committee (HREC 11/QCQ/14). Written informed consent to participate was obtained from each participant. The study design and reporting follow the COnsensus-based Standards for the selection of health status Measurement INstruments (COSMIN) checklist.²²

Table 1.

Participant inclusion and exclusion criteria and description of the study setting and processes.

Inclusion criteria	Presence of at least one chronic disease, aged 18 years or older, attendance at their scheduled clinic intake appointment and a minimum of one intervention session.⁵
Exclusion criteria	Presentation with a severe communication disability or behavioural condition compromising their ability to complete study questionnaires, a cognitive deficit impairing ability to read and write, and/or a known pregnancy up to six months post-partum.⁵
Study setting and processes	The clinic provided training for students as well as for addressing the complex needs of chronic disease patients in the community by providing fully funded services that were tailored to patient needs rather than to conditions. Services were accessed by referral from a general practitioner (GP), local hospital or self-referral.

Procedure

The reproducibility subsample self-completed the DBMA (written in English) and SF-36 at baseline in the clinic environment and at home one week later, in paper form, between April and July 2012. The DBMA did not explicitly state whether a diagnosis was required for a chronic disease to be reported; thus, information reported was based on the judgement of the participant. The longitudinal validity and interpretability sample self-completed the same measures in the clic environment at baseline and at three-month follow-up either in the clinic environment or at home for participants who were going to be unable to attend the clinic at the three-month follow-up.

Outcome measures

The DBMA lists 21 physical conditions (Table 2, first column) and also has an open-ended section where other conditions can be added. Each condition is rated as present or absent. If present, the condition is rated on a 5-point Likert scale (1 = limits daily activities not at all to 5 = limits daily activities a lot), with higher scores indicating greater disease burden. The revised 21-item English version¹⁶ of the original 23-item questionnaire¹ was used for this study. The total score for impact on daily activities is the sum of recorded scores for each condition.^1,¹⁶ The DBMA has no theoretical ceiling as participants can list an unlimited number of ‘other’ conditions in the open-ended section.¹ Non-responses (i.e. no score for the absence of the listed condition and impact on daily activities) were coded as missing values. A single assessor, who was a research assistant for the study with a clinical background in clinical measurements for acute and chronic diseases, administered the questionnaires for the reproducibility component and administered most of the questionnaires for the longitudinal validity and interpretability component, although another two assessors were involved.

Table 2.

Reproducibility of the Disease Burden Morbidity Assessment (DBMA) using a one week retest.

Limitations on daily activities scores for individual items and the total score	No. of paired obs.	Median (IQR) – initial test except where indicated	Median (IQR) – retest except where indicated	ICC (95%CI)	Residual from ICC ANOVA	SEM	SDC	No. (%) of paired observations with exact agreement	No. (%) of paired observations with 1 point difference	No. (%) of paired observations with 2 point difference
High BP	29	1 (1)	1 (2)	0.21 (−0.14 to 0.52)	1.06	1.03	2.86	16 (47.1%)	23 (67.6%)	25 (73.5%)
High cholesterol	28	1 (1)	1 (2)	0.71 (0.46–0.85)	0.37	0.61	1.69	16 (57.1%)	24 (85.7%)	28 (100%)
Asthma	11	2 (2)	2 (3)	0.75 (0.29–0.93)	0.50	0.71	1.97	7 (63.6%)	9 (81.8%)	11 (100%)
Diabetes	17	3 (3)	2 (2)	0.52 (0.05–0.80)	0.90	0.95	2.63	8 (47.1%)	12 (70.6%)	16 (94.1%)
Thyroid disorder	4	1 (1)	3 (3)	–	–	–	–	–	–	–
Chronic back pain or sciatica	23	3 (3)	3 (3)	0.74 (0.49–0.88)	0.56	0.75	2.08	9 (39.1%)	21 (91.3%)	22 (95.7%)
RA	4	4 (3)	3 (2)	–	–	–	–	–	–	–
Rheumatic disease	1	–	–	–	–	–	–	–	–	–
Stomach problem	20	2 (2)	2 (2)	0.36 (−0.11 to 0.69)	1.05	1.02	2.83	6 (30.0%)	13 (65.0%)	19 (95.0%)
Colon problem	7	3 (2)	2 (3)	–	–	–	–	–	–	–
Poor blood circ	15	3 (3)	3 (3)	0.79 (0.48–0.92)	0.42	0.65	1.80	6 (40.0%)	14 (93.3%)	15 (100%)
Overweight	40	3 (3)	3 (2)	0.53 (0.27–0.72)	0.80	0.89	2.47	8 (20.0%)	26 (65.0%)	33 (82.5%)
Hard of hearing	19	2 (2)	2 (1)	0.24 (−0.21 to 0.62)	0.90	0.95	2.63	3 (15.8%)	16 (84.2%)	18 (94.7%)
Vision problem	33	1 (1)	1 (2)	0.28 (−0.7 to 0.57)	0.98	0.99	2.74	14 (43.8%)	25 (78.1%)	32 (100%)
Chronic bronchitis or emphysema	4	3 (2)	4 (4)	–	–	–	–	–	–	–
Osteoarthritis	20	3 (2)	3 (1)	0.42 (<0.01–0.72)	0.96	0.98	2.72	7 (35.0%)	13 (65.0%)	19 (95.0%)
Osteoporosis	3	3 (3)	3 (2)	–	–	–	–	–	–	–
Cancer (last five years)	3	1 (3)	1 (3)	–	–	–	–	–	–	–
CH Failure	0	–	–	–	–	–	–	–	–	–
Heart disease	15	2 (1)	2 (2)	0.05 (−0.51 to 0.55)	1.89	1.37	3.80	3 (20.0%)	6 (40.0%)	12 (80.0%)
Stroke	4	2 (2)	1 (2)	–	–	–	–	–	–	–
No. conditions – mean (SD)	56	7.45 (2.92)^a	6.96 (2.94)^b	N/A	N/A	N/A	N/A	15 (27.0%)	34 (61.0%)	42 (75.0%)
Total impact on daily activities mean (SD)	56	18.02 (10.49)^a	18.11 (12.32)^b	0.77 (0.63–0.86)	1.87	1.37	3.80	3 (5.7%)	11 (21.2% )	16 (30.8%)

Note: Dashes indicate where there were too few cases for a specific disease for values to be calculated.

No.: number; obs.: observations; SD: standard deviation; SEM: standard error of measurement; SDC: smallest detectable change; N/A: not applicable; BP: blood pressure; RA: rheumatoid arthritis; Circ: circulation; CH Failure: congestive heart failure; IQR: interquartile range; ICC: Intraclass correlation coefficient; ANOVA: analysis of variance.

^an = 62.

^bn= 56.

The eight dimensions of the SF-36 were used as an external criterion of health-related quality of life for testing longitudinal validity as per previous studies.^1,17 The SF-36 dimensions were scored on a 0 to 100% scale with 0 indicating the worst function and 100 the best function. The item on self-reported health perception was used as a criterion for determining health stability between baseline and retest as this measure has been found to be a simple, integrative patient-centred assessment for the evaluation of illness in the context of multimorbidity over and above psychosocial measures.²³ The SF-36 has been shown to have acceptable reliability and construct validity in Australian chronic disease populations.^24,25

Statistical analysis

Descriptive statistics were used to describe the study sample. Data were checked regarding whether the assumption of missing completely at random was met using Little’s Completely Missing at Random (CMAR) test. Reproducibility includes both reliability and agreement.²⁶ Reliability of the total score for impact on daily activities and number of comorbidities was assessed using ICCs, using a two-way, random effects model, and associated CIs. An ICC of greater than 0.7 was considered acceptable for research purposes.²¹ Visual inspection of a Bland–Altman plot and linear regression was conducted to determine whether proportional bias was contributing to the ICCs. Agreement was assessed using: percentages of agreement within one and two points between test and re-test on the DBMA; and the standard error of measurement (SEM) and smallest detectable change (SDC) for items with normally distributed residuals. The SEM was calculated as √σ² (where σ² is the mean square error term from the ICC ANOVA).^21,26 The SDC was calculated as 1.96 × √2 × SEM. Interpretability was determined by the percentage of participants who had the lowest and highest total score and the distribution of scores²² at baseline and three-month follow-up, as participants who already have the highest or lowest scores at any point cannot change further in the respective direction impacting on responsiveness, and if large numbers of participants have the same scores they cannot be differentiated from each other influencing reliability.²⁷ Stability as a necessary prerequisite for examining test–retest reliability²⁷ was examined using median or mean individual item and total impact on daily activities scores of the DBMA and the self-reported health perception item of the SF-36. The distribution of scores, mean, and standard deviation of the DBMA sample total scores, median (IQR) of the DBMA individual items, and missing values were also calculated as indicators of interpretability.²⁶

For consistency with previous recommendations, missing SF-36 data were imputed using the mean values of the remaining dimension items²⁸ when less than 50% of the items were missing. When 50% or more of items were missing, data were removed from analyses. Convergent and divergent construct longitudinal validity was determined by examining hypothesised correlations between changes in the DBMA and changes in the SF-36 dimensions using Pearson's product-moment correlation coefficients (or Spearman's correlation coefficients for gender subgroup analyses), determined a priori (Table 3). Internal consistency of the total number of comorbidities and the total impact on daily activities was examined using Chronbach’s alpha. Statistical analyses were performed using IBM SPSS Statistics for Windows, Version 23.0. (Armonk, New York: IBM Corporation).

Table 3.

Hypotheses regarding longitudinal validity and interpretability of the DBMA using correlations with the SF-36 dimensions.^a

Hypotheses	Supported/ partly supported/ not supported	Reasons for being supported/ partly supported or not supported
(1) Correlations between changes in the DBMA impact of respiratory conditions, stroke, diabetes, cardiovascular conditions, and arthritis on daily activities with changes in SF-36 physical functioning dimensions were expected to be stronger than the same correlations for other conditions based on findings across studies that patients with those conditions have lower functioning in overall HRQOL or physical functioning dimensions than people with other conditions from cross-sectional studies^28–30 and that congestive heart failure, diabetes, and chronic respiratory conditions predict clinically significant decline in SF-36 physical components scores.¹⁴	Partly supported	Correlations were also similarly strong for chronic back pain and osteoporosis. Different classifications of cardiovascular conditions in previous studies meant conditions such as high blood pressure were not previously investigated thus were unable to be included in the hypothesis.
(2) Correlations between changes in the DBMA impact on daily activities with changes in the SF-36 physical functioning, bodily pain,^1,31 and role physical¹ dimensions were expected to be greater than correlations of change in the DBMA impact on daily activities with other SF-36 dimensions.	Partly supported	While correlations were generally stronger for the physical function, role physical, and bodily pain dimensions correlations were also strong for the role emotional dimension based on the number of items with the two strongest correlations across dimensions.
(3) For conditions where pain was likely to be a strong feature of the condition such as back pain or sciatica, osteoarthritis, osteoporosis, rheumatoid arthritis, and overweight,³¹ the correlations between changes in the DBMA impact on daily activities and changes in the SF-36 bodily pain dimension were expected to be negative and stronger than other correlations.	Supported	Correlations between changes in the DMBA impact on daily activities and changes in the SF-36 bodily pain dimension were negative and generally stronger for these conditions than correlations between these respective variables for other conditions. However, unexpectedly there were a number of other conditions for which there were strong correlations also such as high blood pressureand hard of hearing.
(4) Changes in SF-36 physical and mental functioning dimensions were expected to be very similar for males and females. This was based on the finding that decrements in the SF-36 physical components score were identical for males and females but decrements in the mental components score were slightly larger for males in five physical disease clusters studied (gastrointestinal, musculoskeletal, respiratory, endocrine, cardiovascular).³²	Partly Supported	Correlations between changes in the DMBA impact on daily activities and changes in the SF-36 physical functioning and mental health functioning dimensions were generally similar for males and females with the difference in Spearman’s correlation coefficients usually less than or equal to 0.20 (data not shown).
(5) The majority of hypothesised stronger correlations of changes in the DMBA impact on daily activities with changes in the SF-36 dimensions were expected to be negative due to scaling for improvement being in the opposite direction on the measures.	Supported	The majority of hypothesised stronger correlations were negative.
(6) Correlations between changes in the DMBA number of conditions and the SF-36 dimensions were expected to be lower than correlations between the DMBA impact of comorbidities on daily activities and the SF-36 dimensions, but stronger for the physical functioning dimensions than mental functioning dimensions based on previous cross-sectional studies of chronic disease and multimorbidity.^1,3,7,16 For example, stronger correlations between the DMBA impact of conditions on daily activities and SF-36 physical functioning (−0.63) have been identified than between the DMBA number of conditions and SF-36 physical functioning (−0.43).¹	Supported	Generally, stronger correlations were found between changes in the number of conditions and changes in the physical functioning dimensions in comparison to mental health dimensions.

^aHypotheses for longitudinal validity were based on expected correlations between changes in SF-36 dimensions with changes in the DMBA total scores and individual items. Hypothesis testing was based on studies of the DBMA wherever possible, as different measures of multimorbidity have beeen found to result in varying levels of physical health-related quality of life.³⁰ As few studies of temporal patterns exist for people with multimorbidity expected changes were based on cross-sectional as well as longitudinal studies and expert opinion.

Results

A summary of the participant flow for the study is presented in Figure 1. The assumption of data missing completely at random for the number of comorbidities and severity of comorbidities was supported for the DBMA (Little’s MCAR test = 0.07 to 0.95). Internal consistency of the number of comorbidities score was 0.60 and of the total impact on daily activities score was 0.69.

Reproducibility component

Of the participants who consented to participation (n = 62 of 78 approached), 90% participated. No data were missing for participants who completed the DBMA at baseline or one-week retest. Participants who completed the initial test and a one-week retest (n = 56) had a mean (SD) age of 61 (12) years, and the majority (n = 34, 61%) were female. The mean (SD) number of comorbidities was 7.5 (2.9), and all participants had two or more comorbidities.

For the total score of the impact of all conditions on daily activities, test–retest reliability was acceptable (ICC = 0.77) (Table 2). The SDC varied from 1.69 to 3.80 on individual items. The reliability of the total number of chronic conditions (Cohen’s kappa = 0.27) indicated this index was not as stable as the total impact on daily activities over the test–retest period (Table 2). A Bland–Altman plot of the total number of chronic conditions and accompanying regression analysis (Figure 2) indicated proportional bias did not appear to be present. The majority of individual items and the total impact on daily activities median and mean scores were relatively stable (i.e. no change for 15 out of 19 relevant individual items, mean (SD) total impact on daily activities score 18.02 (10.49) at baseline and 18.11 (12.32) at retest). Self-reported health perception was relatively stable between baseline and retest (mean (SD) = 3.61 (0.70), 3.55 (0.79), respectively) based on the patients who completed the health perception item of the SF-36 at both timepoints.

Longitudinal validity and interpretability component

Participants (n = 307/351, 87%) who completed the DBMA at baseline and at three-month follow-up (Figure 1) had a mean (SD) age of 59 (14) years and the majority (n = 178, 58%) were female. DBMA missing data at baseline for 10 participants are reported in Table 4. No DBMA data were missing for participants who completed the DBMA at three-month follow-up. The mean (SD) number of comorbidities at baseline was 7.0 (3.0), and n = 346 (99%) had two or more comorbidities. The most common condition at baseline was being overweight (76%). Ratings of the impact on daily activities (Table 4) were skewed with the impact of individual conditions in the lower range for the majority of participants, with median ratings across the 21 listed conditions of 3 or less (on the 5-point scale) for all but the rheumatology item (median 4 at baseline).

Table 4.

Disease burden morbidity assessment (DBMA) medians and interquartile ranges at baseline and three-month follow-up (longitudinal validity study component).

Limitations on daily activities rating by specific condition or total score and total number of conditions^a,b	Number (%) of participants at baseline	Number (%) of participants at follow-up	Median (IQR) at baseline except where indicated	Median (IQR) at three-month follow-up except where indicated
High blood pressure	200 (57%)	167 (54%)	1.0 (1.0)	1.0 (1.0)
High cholesterol	187 (53%)	153 (50%)	1.0 (1.0)	1.0 (1.0)
Asthma	109 (31%)	84 (27%)	2.0 (2.0)	2.0 (2.0)
Diabetes	149 (43%)	130 (42%)	2.0 (3.0)	2.0 (2.0)
Thyroid	44 (13%)	34 (11%)	2.0 (2.0)	1.0 (1.0)
Chronic back pain or sciatica	189 (55%)	174 (57%)	3.0 (3.0)	3.0 (2.0)
Rheumatoid arthritis	36 (10%)	37 (12%)	3.0 (2.0)	3.0 (1.0)
Rheumatic disease	12 (3%)	11 (4%)	4.0 (3.0)	1.0 (3.0)
Stomach problem	137 (39%)	101 (33%)	2.0 (2.0)	2.0 (2.0)
Colon problem	72 (21%)	53 (17%)	3.0 (3.0)	2.0 (2.0)
Poor blood circulation	124 (35%)	104 (34%)	3.0 (2.0)	2.0 (1.0)
Overweight	267 (76%)	233 (76%)	3.0 (3.0)	3.0 (2.0)
Hard of hearing	138 (39%)	127 (41%)	2.0 (2.0)	2.0 (2.0)
Vision problem	255 (73%)	201 (66%)	2.0 (2.0)	2.0 (2.0)
Chronic bronchitis or emphysema	39 (11%)	29 (9%)	3.0 (2.0)	3.0 (3.0)
Osteoarthritis	175 (50%)	154 (50%)	3.0 (2.0)	3.0 (2.0)
Osteoporosis	48 (14%)	42 (14%)	2.0 (3.0)	2.0 (2.0)
Cancer during the last 5-yrs	18 (5%)	13 (4%)	2.5 (2.5)	1.0 (3.5)
Congestive heart failure	13 (4%)	15 (5%)	3.0 (2.5)	2.0 (2.0)
Heart disease	74 (21%)	65 (21%)	3.0 (2.0)	2.0 (2.0)
Stroke	29 (8%)	19 (6%)	2.0 (2.5)	1.0 (2.0)
Total number of conditions mean (SD)	351	307	7.0 (3.0)	6.6 (2.8)
Total limitation on daily activities mean (SD)	351	307	17.7 (10.5)	15.9 (8.9)

IQR: interquartile range.

^aConditions listed by participants in the “other” section of the DBMA but not included in the table were depression, post-polio syndrome, epilepsy, anxiety, schizophrenia, neck pain, vertigo, obstructive sleep apnoea, hemochromatosis, non-alcoholic fatty liver disease, ross river fever, spinal fusion, diverticulitis, Reynaud’s syndrome, benign intercranial hypertension, glaucoma, gout, tinnitus, carpal tunnel, and kidney disease. A further rare condition was documented by a participant but has not been specified to prevent any risk of the patient being identified.

^bDBMA missing data at baseline included back data for four participants, thyroid and back problems for one participant, rheumatoid arthritis for one participant, vision problem and chronic bronchitis or emphysema for one participant, osteoporosis for one participant, vision and hearing problems for one participant, diabetes for two participant, asthma for one participant.

The strength and direction of correlations between changes in SF-36 dimensions with DBMA total scores and individual items are displayed in Table 5. The longitudinal validity hypotheses were supported or partly supported (Table 3), although some unexpected higher strength correlations were observed for conditions in three of the six hypotheses. With respect to interpretability, a low percentage of participants had the lowest possible total number of comorbidities at baseline and three-month follow-up (1.4 and 3.9%, respectively) and for the total score of impact on daily activities at baseline and three-month follow-up (0.6 and 1.3%, respectively) indicating floor effects were unlikely to have impacted on estimates. Analyses pertaining to the hypothesis regarding the subgroups of males versus females were likely adequately powered as data for greater than 50 participants were available for each subgroup¹⁹ (n = 178 females, n = 129 males).

Table 5.

Correlations between changes from baseline to three months on the disease burden morbidity assessment (DBMA) and changes in the 36-item Short Form Health Survey SF-36 physical and mental functioning dimensions (longitudinal validity study component).

DBMA items (number of participants with data included in correlations)^a	Change in the SF-36 physical functioning dimensions^b				Change in the SF-36 mental functioning dimensions^b
	Physical functioning	Role physical	Bodily pain	General health	Vitality	Social functioning	Role emotional	Mental health
High blood pressure (n = 151-154)	−0.11	−0.10	−0.12	−0.03	−0.05	−0.10	0.05	−0.10
High cholesterol (n = 131-133)	−0.12	−0.07	<0.01	−0.07	−0.04	−0.12	0.11	−0.06
Asthma (n = 71–73)	−0.21	−0.24*	−0.13	−0.36**	−0.34**	−0.10	0.05	−0.13
Diabetes (n = 123-124)	−0.28**	−0.20*	−0.13	−0.23*	−0.24**	−0.13	−0.10	−0.10
Thyroid disorder (n = 30)	−0.07	−0.04	−0.27	0.15	0.21	0.06	0.42*	0.16
Chronic back pain or sciatica (n = 140-141)	−0.31**	−0.04	−0.35**	−0.15	−0.16	−0.21*	0.03	−0.09
Rheumatoid arthritis (n = 20)	−0.08	−0.02	−0.14	0.18	−0.16	−0.07	0.42	0.04
Stomach problem (n = 88–89)	−0.21	0.10	0.05	0.08	−0.06	−0.19	0.13	−0.24*
Colon problem (n = 39–40)	0.03	−0.27	−0.18	−0.11	−0.15	−0.03	−0.24	−0.09
Poor blood circulation (n = 71)	−0.06	0.06	−0.25*	−0.08	<−0.01	−0.19	−0.20	−0.22
Overweight (n = 219–222)	−0.18**	0.04	−0.19**	−0.11	−0.07	−0.16*	0.05	−0.06
Hard of hearing (n = 104–105)	−0.12	−0.08	−0.24*	−0.27**	−0.20*	−0.20*	−0.08	−0.07
Vision problem (n = 179–181)	−0.13	−0.03	−0.04	−0.16*	−0.08	−0.07	0.02	0.01
Chronic bronchitis or emphysema (n = 21–22)	−0.01	−0.33	0.01	−0.25	0.01	0.07	−0.33	−0.08
Osteoarthritis (n = 129–130)	−0.12	0.09	−0.25**	−0.08	−0.02	0.10	0.08	0.09
Osteoporosis (n = 28–29)	−0.31	0.10	−0.20	−0.29	0.14	<0.01	0.40*	0.09
Heart disease (n = 53–54)	−0.33*	−0.24	−0.22	0.03	−0.26	−0.20	−0.25	−0.13
Total score for limitations on daily activities
(n = 303–306)	−0.21**	−0.09	−0.26**	−0.21**	−0.13*	−0.18**	0.03	−0.06
Total score for number of comorbidities
(n = 303–306)	−0.06	<0.01	−0.16**	−0.07	0.04	0.04	0.06	<0.01

Note: 36-item Short Form Health Survey: SF-36.^aCorrelations have only been calculated for items with greater than n = 20.

^bNo SF-36 data missing at baseline for participants who were included at baseline and three-month follow-up, one participant had all SF-36 data missing at three-month follow-up, three participants had data for two dimensions missing at three-month follow-up, and one participants had data for one dimension missing at three-month follow-up.

*p < 0.05, **p < 0.01.

Discussion

This study indicated that the DBMA had acceptable psychometric properties for evaluating disease burden in people aged 18 years or older with chronic disease and multimorbidity in several areas (e.g. longitudinal validity of individual items and test–retest reliability of the total score of the impact on daily living). The sample used to test the DBMA was unique in that individuals with a large number of conditions were included, which has not been the focus of previous work. Interpretability was supported based on the small percentage of participants with the lowest scores.

Variability in the impact of some diseases on daily activities was likely responsible for the low reliability coefficients for these items (i.e. diabetes, stomach problem, vision problem, osteoarthritis) rather than measurement error. Under-reporting of stigmatizing chronic diseases and variability in patient-reports of conditions with highly subjective symptoms such as pain in arthritis and headache⁶ may have impacted the findings in this study. It was also plausible that variability of the impact of individual conditions such as overweight and hard of hearing was due to variability in the impact of other comorbid conditions that was difficult to isolate or that ratings across multiple items were influenced by dominant conditions at that time. For example, separating out the impact of pain underlying osteoarthritis from the impact of pain underlying cancer might be difficult for patients with both conditions. This reasoning is potentially supported by the unexpected relatively high correlations between changes in the SF-36 bodily pain dimension and the impact of conditions that have not typically been conceptualised as pain-centric conditions (colon problem, blood circulation problem, overweight, hard of hearing). Short-term fluctuations in disease burden attributable to individual conditions pose an important challenge to those seeking to quantify disease burden among people with chronic conditions; however, findings from this study indicated that the DBMA total score of the impact on daily living was acceptable for picking up changes in the study sample.

Interestingly, the reliability of using a total count of conditions using the DBMA was not supported in people with a large mean number of chronic conditions in this study. Counts of the number of chronic conditions have been important when quantifying multimorbidity in prior studies.⁴ Acceptable agreement of more than 90% for all but one item has been reported for a patient-report comorbidity measure based on the Charlson index in an inpatient sample, although only 3 out of 18 individual items had test–retest reliability coefficients above 0.70.³³ Variability in the total number of conditions reported by chronic disease patients over the one-week period in this study may be related to inaccurate memory of conditions (recall bias),³⁴ and variability in the setting or in reporting due to short-term fluctuations in symptoms (e.g. arthritis may have been reported on days where stiffness was felt by patients even if there was no formal diagnosis of arthritis³⁵). The reliability results using the total score of impact on daily activities were consistent with results of testing using the DBMA-Fr, albeit that our test–retest reliability results were marginally lower than in that study (ICC = 0.77 vs. 0.86¹⁸).

Limitations and future directions

Support exists regarding the validity of summing scores of the impact of single conditions on daily activities in a population sample of adults 50 years or older with four or fewer comorbidities.³⁶ Further, new work that applied multi-trait multi-method analysis has supported the validity of using a summed disease burden score for a disease burden measure with multimorbid conditions using a list of 35 conditions.³⁷ However, further work is recommended to confirm the use of DBMA total scores. Although the impact on daily activities may be apportioned across conditions in cases where people find it difficult to attribute symptoms to individual diseases,³⁸ the potential difficulty that patients face in judging how much impact they apportion to each condition may be considered a limitation of the DBMA. Similarly, patients may inadvertently attribute the impact of multiple chronic conditions to a single condition which may complicate disease burden estimates.³⁵ However, it is noteworthy that in both of these scenarios, the validity of the total score of the impact of individual conditions would be supported.

The focus of the DBMA was on physical conditions. Mental health conditions could be reported in the ‘other’ open-ended section of the measure but should be measured separately 1,18 for those seeking to capture the impact of all comorbidities as the prevalence of these conditions was likely under-reported. Additionally, the relatively small sample size for testing reproducibility for some individual items may have meant those item findings were influenced by unusual reporting behaviour from a small number of participants, thus further work is required to confirm (or refute) some of this study’s findings. Further content validation is required that involves target groups and examines temporal trends. A further consideration is that the DBMA may perform differently in people with few comorbidities as mean number of comorbidities in the study sample was seven, thus further testing should be conducted with this group.

Footnotes

References

Subjective assessments of comorbidity correlate with quality of life health outcomes: initial validation of a comorbidity assessment instrument. Health Qual Life Outcomes 2005; 3: 51

Pietersma

de Vries

and van den Akker-van ME. Domains of quality of life: results of a three-stage Delphi consensus procedure among patients, family of patients, clinicians, scientists and the general public. Qual Life Res 2014; 23: 1543–1556.

Byles

D'Este

Parkinson

et al . Single index of multimorbidity did not predict multiple outcomes. J Clin Epidemiol 2005; 58: 997–1005.

McPhail

SM.

Multimorbidity in chronic disease: impact on health care resources and costs. Risk Manag Healthc Policy 2016; 9: 143–156

Tyack

Frakes

Cornwell

et al . Predictors of health-related quality of life in people with a complex chronic disease including multimorbidity: a longitudinal cohort study. Qual Life Res 2016; 25: 2579–2572.

van der Velden

What do health interview surveys tell us about the prevalences of somatic chronic diseases? A study into concurrent validity. Eur J Public Health 1998; 8: 52–58.

et al . Comparative assessment of three different indices of multimorbidity for studies on health-related quality of life. Health Qual Life Outcomes 2005; 3: 74.

Diederichs

Wellmann

Bartels

et al . How to weight chronic diseases in multimorbidity indices? Development of a new method on the basis of individual data from five population-based studies. J Clin Epidemiol 2012; 65: 679–685.

Tooth

Hockey

Byles

et al . Weighted multimorbidity indexes predicted mortality, health service use, and health-related quality of life in older women. J Clin Epidemiol 2008; 61: 151–159.

et al . Why summary comorbidity measures such as the Charlson comorbidity index and Elixhauser score work. Med Care 2015; 53: e65–e72.

et al . Measures of multimorbidity and morbidity burden for use in primary care and community settings: a systematic review and guide. Ann Fam Med 2012; 10: 134–141.

et al . Defining comorbidity: implications for understanding health and health services. Ann Fam Med 2009; 7: 357–363.

Comorbidity or multimorbidity. Eur J Gen Pract 1996; 2: 65–70.

14.

Bayliss

Ware

et al . Predicting declines in physical function in persons with multiple medical conditions: What can we learn from the medical problem list. Health Qual Life Outcomes 2004; 2: 47.

et al . Cumulative complexity: a functional, patient-centered model of patient complexity can improve research and practice. J Clin Epidemiol 2012; 65: 1041–1051.

Seniors' self-reported multimorbidity captured biopsychosocial factors not incorporated into two other data-based morbidity measures. J Clin Epidemiol 2009; 62: 550–557.

et al . Relationship between multimorbidity and health-related quality of life of patients in primary care. Qual Life Res 2006; 15: 83–91.

et al . Validation of the disease burden morbidity assessment by self-report in a French-speaking population. BMC Health Serv Res 2012; 12: 1–6.

et al . On assessing responsiveness of health-related quality of life instruments: guidelines for instrument evaluation. Qual Life Res 2003; 12: 349–362.

Planning a reproducibility study: how many subjects and how many replicates per subject for an expected width of 95 percent confidence interval for the intraclass correlation coefficient?

Stat Med 2001; 20: 3205–3214.

et al . Quality criteria were proposed for measurement properties of health status questionnaires. J Clin Epidemiol 2007; 60: 34–42.

et al . The COSMIN checklist for assessing the methodological quality of studies on measurement properties of health status measurement instruments: an international Delphi study. Qual Life Res 2010; 19: 539–549.

et al . Association of self-rated health with multimorbidity, chronic disease and psychosocial factors in a large middle-aged and older cohort from general practice: a cross-sectional study. BMC Fam Pract 2014; 15: 185.

24.

Sanson-Fisher

RW.

Adaptation and Validation of the SF-36 Health Survey for use in Australia. J Clin Epidemiol 1998; 51: 961–967.

25.

McCallum

The SF-36 in an Australian sample: validating a new, generic health status measure. Aust J Public Health 1995; 19: 160–166.

et al . When to use agreement versus reliability measures. J Clin Epidemiol 2006; 59: 1033–1039.

et al . The COSMIN checklist manual, www.cosmin.nl (accessed 6 July 2017).

28.

Ware

Snow

Kosinski

et al.

SF-36 health survey. Manual and interpretation guide. Boston: The Health Institute, New England Medical Center, 1993.

29.

Wang

Beyer

Gensichen

et al . Health-related quality of life among general practice patients with differing chronic diseases in Germany: cross sectional survey. BMC Public Health 2008; 8: 246.

30.

Saarni

Harkanen

Sintonen

et al . The impact of 29 chronic conditions on health-related quality of life: a general population survey in Finland using 15D and EQ-5D. Qual Life Res 2006; 15: 1403–1414.

31.

Kriegsman

DMW

Deeg

DJH

Stalman

WAB.

Comorbidity of somatic chronic diseases and decline in physical functioning: the longitudinal aging study Amsterdam. J Clin Epidemiol 2004; 57: 55–65.

32.

Brettschneider

Leicht

Bickel

et al . Relative impact of multimorbid chronic conditions on health-related quality of life – results from the MultiCare Cohort Study. PLoS One 2013; 8: e66742

33.

Katz

Chang

Sangha

et al . Can comorbidity be measured by questionnaire rather than medical record review? Med Care 1996; 34: 73–84.

34.

McPhail

Haines

Response shift, recall bias and their effect on measuring change in health-related quality of life amongst older hospital patients. Health Qual Life Outcomes 2010; 8: 65.

35.

Kriegsman

Penninx BW, Van Eijk

et al . Self-reports and general practitioner information on the presence of chronic diseases in community dwelling elderly: a study on the accuracy of patients' self-reports and on determinants of inaccuracy. J Clin Epidemiol 1996; 49: 1407–1417.

36.

Agborsangaya

Lau

Lahtinen

et al . Health-related quality of life and healthcare utilization in multimorbidity: results of a cross-sectional survey. Qual Life Res 2013; 22: 791–799.

37.

Ware

Jr. Gandek

Allison

The validity of disease-specific quality of life attributions among adults with multiple chronic conditions. Int J Stats Med Res 2016; 5: 17–40.

38.

McDaid

Hanly

Richardson

et al. The effect of multiple chronic conditions on self-rated health, disability and quality of life among the older populations of Northern Ireland and the Republic of Ireland: a comparison of two nationally representative cross-sectional surveys. BMJ Open 2013; 3: e002571.

Reproducibility,longitudinal validity and interpretability of the Disease Burden Morbidity Assessment in people with chronic disease

Abstract

Objective

Methods

Results

Discussion

Keywords

Introduction

Methods

Study design and participants

Procedure

Outcome measures

Statistical analysis

Results

Reproducibility component

Longitudinal validity and interpretability component

Discussion

Limitations and future directions

Footnotes

Acknowledgements

Declaration of conflicting interests

Funding

References