The impact of varying the number and selection of conditions on estimated multimorbidity prevalence: A cross-sectional study using a large, primary care population dataset

Clare MacRae; Megan McMinn; Stewart W Mercer; David Henderson; David A McAllister; Iris Ho; Emily Jefferson; Daniel R Morales; Jane Lyons; Ronan A Lyons; Chris Dibben; Bruce Guthrie

doi:10.1371/journal.pmed.1004208

. 2023 Apr 4;20(4):e1004208. doi: 10.1371/journal.pmed.1004208

The impact of varying the number and selection of conditions on estimated multimorbidity prevalence: A cross-sectional study using a large, primary care population dataset

Clare MacRae ^1,^2,^*, Megan McMinn ², Stewart W Mercer ^1,², David Henderson ², David A McAllister ³, Iris Ho ^1,², Emily Jefferson ⁴, Daniel R Morales ⁴, Jane Lyons ⁵, Ronan A Lyons ⁵, Chris Dibben ⁶, Bruce Guthrie ^1,²

PMCID: PMC10072475 PMID: 37014910

Abstract

Background

Multimorbidity prevalence rates vary considerably depending on the conditions considered in the morbidity count, but there is no standardised approach to the number or selection of conditions to include.

Methods and findings

We conducted a cross-sectional study using English primary care data for 1,168,260 participants who were all people alive and permanently registered with 149 included general practices. Outcome measures of the study were prevalence estimates of multimorbidity (defined as ≥2 conditions) when varying the number and selection of conditions considered for 80 conditions. Included conditions featured in ≥1 of the 9 published lists of conditions examined in the study and/or phenotyping algorithms in the Health Data Research UK (HDR-UK) Phenotype Library. First, multimorbidity prevalence was calculated when considering the individually most common 2 conditions, 3 conditions, etc., up to 80 conditions. Second, prevalence was calculated using 9 condition-lists from published studies. Analyses were stratified by dependent variables age, socioeconomic position, and sex. Prevalence when only the 2 commonest conditions were considered was 4.6% (95% CI [4.6, 4.6] p < 0.001), rising to 29.5% (95% CI [29.5, 29.6] p < 0.001) considering the 10 commonest, 35.2% (95% CI [35.1, 35.3] p < 0.001) considering the 20 commonest, and 40.5% (95% CI [40.4, 40.6] p < 0.001) when considering all 80 conditions. The threshold number of conditions at which multimorbidity prevalence was >99% of that measured when considering all 80 conditions was 52 for the whole population but was lower in older people (29 in >80 years) and higher in younger people (71 in 0- to 9-year-olds). Nine published condition-lists were examined; these were either recommended for measuring multimorbidity, used in previous highly cited studies of multimorbidity prevalence, or widely applied measures of “comorbidity.” Multimorbidity prevalence using these lists varied from 11.1% to 36.4%. A limitation of the study is that conditions were not always replicated using the same ascertainment rules as previous studies to improve comparability across condition-lists, but this highlights further variability in prevalence estimates across studies.

Conclusions

In this study, we observed that varying the number and selection of conditions results in very large differences in multimorbidity prevalence, and different numbers of conditions are needed to reach ceiling rates of multimorbidity prevalence in certain groups of people. These findings imply that there is a need for a standardised approach to defining multimorbidity, and to facilitate this, researchers can use existing condition-lists associated with highest multimorbidity prevalence.

Using a UK wide primary-care dataset and including more than 80 index conditions, Clare MacRae and colleagues report multimorbidity prevalence and how prevalence estimates change when varying the number and type of index conditions.

Author summary

Why was this study done?

There is wide variety in the conditions considered by researchers when measuring multimorbidity prevalence.
A systematic review of 566 studies, published in 2021, found lack of consensus in the selection of conditions considered.
In half of studies only 8 conditions (diabetes, stroke, cancer, chronic obstructive pulmonary disease, hypertension, coronary heart disease, chronic kidney disease, and heart failure) were consistently considered, and the number of conditions considered varied from 2 to 285 (median 17).
A more consistent approach to measuring multimorbidity is needed to facilitate comparability and generalisability across studies.

What did the researchers do and find?

We examined the impact of varying the conditions considered when measuring multimorbidity prevalence. We combined different numbers of conditions (from a list of 80) and selections of conditions (using 9 published condition-lists used to define and measure comorbidity, multimorbidity, and its prevalence) to determine how multimorbidity prevalence changed. All conditions were counted in the same way using publicly available code lists.
There are large differences in prevalence, a range of 4.6% to 40.5%, when different numbers and selections of conditions are considered.
People who are the oldest, living in the most deprived areas, and men require fewer conditions to be considered to reach close to multimorbidity prevalence when considering all 80 conditions (the ceiling effect, where the prevalence approaches the upper limit of prevalence possible in the study).
Highest multimorbidity prevalence was found when using the Ho always + usually (derived from a recent Delphi consensus study), Barnett (widely used to measure multimorbidity prevalence), and Fortin (recommended for use in measuring multimorbidity) condition-lists.

What do these findings mean?

There is a need for standardisation when measuring multimorbidity prevalence so that results across studies are comparable and population subgroups are accurately represented.
To address this, researchers can consider using the Ho always + usually, Barnett, or Fortin condition-lists that report the highest and most stable estimates of multimorbidity prevalence (where adding further conditions to the count had very little impact).

Introduction

Multimorbidity is increasing in prevalence due to improved survival from chronic diseases and population ageing, and now poses major challenges to healthcare systems worldwide [1]. Multimorbidity is common, increases substantially with advancing age, and is more common in women and people with lower socioeconomic position (SEP) [2,3]. Despite its importance, existing research literature is highly heterogenous in how it defines and measures multimorbidity [4]. Choice of conditions considered in the count (the denominator) when measuring multimorbidity prevalence is likely to be driven by pragmatic decision-making in the context of data availability [5], or by recycling of existing published condition-lists, and results in wide diversity in the number and selection of conditions considered in current multimorbidity literature [6]. In a systematic review of 566 studies of multimorbidity, the number of conditions considered in counts of multimorbidity prevalence ranged from 2 to 285 (median 17, interquartile range [IQR] 11 to 23) [4]. Only 8 core conditions were consistently considered in more than half of studies (diabetes, stroke, cancer, chronic obstructive pulmonary disease, hypertension, coronary heart disease, chronic kidney disease, and heart failure) [4].

As a result of this diversity, multimorbidity prevalence estimates vary widely across studies [7], making it difficult to make comparisons within the existing literature. Unsurprisingly, higher multimorbidity prevalence is reported by studies that consider a larger number of conditions in their count of multimorbidity [7,8], studies that consider conditions that are most prevalent [2], and in studies that include more people in older age groups [7]. The number and selection of conditions considered when measuring multimorbidity prevalence is therefore important, but there is little consistent guidance to support researchers when deciding which conditions to consider. Researchers have recommended condition-lists to consider in multimorbidity measurement, including 11 conditions by Diederichs and colleagues [6] and 20 conditions by Fortin and colleagues [9]. More recently, a modified Delphi panel study by Ho and colleagues [10] developed 2 condition-lists based on international consensus on the measurement of multimorbidity: one list recommending conditions to always consider and a second recommending conditions to usually consider when counting multimorbidity prevalence [10].

All multimorbidity research findings are dependent on decisions made at the earliest stages in measurement, including what has been measured, and therefore, building understanding of the properties of multimorbidity as a concept is needed. To inform researchers’ choices, and improve the comparability and reproducibility of future research, it is important to understand the relationship between multimorbidity prevalence and the number and selection of conditions considered in the count. The aim of this study was to examine these relationships in a large primary care cohort.

Methods

Study design

A cross-sectional study design was used to examine the hypothesis that multimorbidity prevalence varies when considering different numbers of conditions, and different selections of conditions (using published recommended or commonly used condition-lists), in the count. The analyses were designed in November 2021, performed in February 2022, and no data-driven changes to analyses took place during this period. As part of the peer-review process, 1 additional condition-list was added [11], and the paper updated to include p-values as well as confidence intervals for proportions, and sensitivity analyses of variation in prevalence by deprivation and sex using raw rather than direct age-standardised data were added.

Data sources

The study analysed routinely collected, anonymised individual-level data from English participants in the Clinical Practice Research Datalink (CPRD) Gold dataset, which are broadly representative of the United Kingdom population [12]. Available data included individual demography (age, SEP, and sex), clinical codes from both GP electronic health records (Read codes) and hospital admission data (ICD-10 codes), and laboratory results. SEP was defined as deciles of the English Index of Multiple Deprivation (IMD), a measure of relative deprivation according to small local area level, with deciles defined by national thresholds [13].

Study participants

Study participants were all people who were alive and permanently registered with 149 included general practices on the study index date, November 30, 2015, with least 2 year’s GP registration prior to this [14].

Definition of variables

For each individual, we defined the presence of 80 conditions using a set of existing code lists that combined Read codes (version 2) applied to GP electronic health record data, International Classification of Diseases 10th version (ICD-10) codes applied to hospital admission data, and laboratory results recorded in the GP electronic health record [15] (S1 Table). The 80 conditions were chosen because they featured in 1 or more of the 9 published lists of conditions examined in the study and/or phenotyping algorithms (condition code lists) in the Health Data Research UK (HDR-UK) Phenotype Library [15]. New code lists were made by study authors for conditions featured in existing condition-lists where no HDR-UK algorithms were available and are listed in S1 Table. All the codes used to identify individuals with each condition were mutually exclusive; therefore, double counting of conditions was not possible, and all conditions contained within condition-lists were included in the total 80 conditions.

Condition-lists were either specifically recommended for measuring multimorbidity (referred to hereinafter as Diederichs [6], Fortin [9], Ho always [10], Ho always + usually [10], and N’Goran [11]), used in previous highly cited large-scale studies of multimorbidity prevalence (Barnett [16] and Salive [17]), or included in widely applied measures of “comorbidity” (Charlson [18] and Elixhauser [19]). The 2 condition-lists recommended by the recent Ho and colleagues Delphi consensus study [10] were conditions recommended to always include (Ho always), and all the conditions recommended by both the lists, conditions to always include and to usually include (Ho always + usually) (S1 Information Panel).

Heterogeneity existed in the description and the hierarchical level of conditions between condition-lists. Therefore, to ascertain each condition in the same way for every condition-list, some were dis-aggregated to more granular descriptions. For example, Diederichs and colleagues [6] considered cancer, while Ho always [10] considered 3 condition groups that were all cancers: primary malignancy, secondary malignancy, and haematological malignancy. In this case, in the Diederichs and colleagues [6] condition-list, cancer was disaggregated from 1 to 3 conditions (to primary malignancy, secondary malignancy, and haematological malignancy) to allow direct comparison with Ho always [10]. Therefore, the number of included conditions in some condition-lists varied from the original published lists (S2 Table).

Statistical analysis

Multimorbidity prevalence was calculated when different numbers and selections of conditions were considered in the count (the denominator). In all analyses, multimorbidity was defined by the cut-off (the numerator) that was the presence of ≥2 conditions [3].

We conducted a suite of comparisons including examination of the effect of the number of conditions considered in the count on multimorbidity prevalence by considering the most common 2 conditions, followed by the most common 3 conditions, the most common 4 conditions, etc., for every number up to considering all 80 conditions in the count (S1 Information Panel). To do this, conditions were ordered from most to least prevalent and added in turn to each new count. The prop.test procedure in R was used to estimate 95% confidence intervals and ps for prevalence were calculated using the normal approximation for large samples.

By making the assumption that multimorbidity prevalence estimated by considering all 80 conditions in the count was the true prevalence, we then calculated the number of conditions that had to be included to exceed a relative risk (RR) of 0.99 of this “true” prevalence. This was done to estimate when a ceiling effect was present, where the prevalence approaches the upper limit of possible prevalence in the study and the point at which adding further conditions to the count had very little impact on multimorbidity prevalence.

To examine the effect the selection of conditions considered in the count, multimorbidity prevalence was calculated when considering the conditions included in each of the 9 condition-lists. Since age is very strongly associated with multimorbidity and the SEP and sex composition within different age groups varies making age a major confounder, analyses were standardised to the age structure of the whole study cohort as the standard population and age-specific standardised rates for population subgroups were calculated [20]. Sensitivity analysis was done using unstandardised rates.

This study adhered to the REporting of studies Conducted using Observational Routinely collected Data reporting guidelines [21] (S3 Table). All data management, statistical analyses, and plotting was done in R version 3.6.2 [22], available within in the ISO27001 and Scottish Government approved Health Informatics Centre Safe Haven. The analysis was approved by CPRD Independent Scientific Advisory Committee (reference 20_018) under the terms of CPRD NHS Research Ethics dataset approval.

Results

The study included 1,168,620 people. When considering all 80 conditions in the count, multimorbidity was present in 473,533 (40.5%) of the cohort. People with multimorbidity were older than the whole population, median 60 years (IQR 46 to 72) versus 44 years (IQR 23 to 60), more often women, 257,237 (54.3%) versus 587,687 (50.3%), and more often lived in the 5 most deprived IMD decile areas, 208,386 (44.0%) versus 505,322 (43.2%). Differences between the whole population and people with multimorbidity examined using χ² tests for proportions within each age group, IMD decile, and both sexes, were statistically significant (p < 0.001) (Table 1).

Table 1. Population characteristics.

	No. (% of total population/each column) N = 1,168,620	No. (% of each row) with multimorbidity when considering all 80 conditions in the count
Age group, years^*
0–9	113,955 (9.8)	2,739 (2.4)
10–19	137,517 (11.8)	9,129 (6.6)
20–29	122,237 (10.5)	24,916 (20.4)
30–39	143,243 (12.3)	39,888 (27.8)
40–49	176,061 (15.1)	66,878 (37.9)
50–59	173,435 (14.8)	89,586 (51.5)
60–69	141,041 (12.1)	98,512 (69.9)
70–79	97,843 (8.4)	82,752 (84.4)
≥80	63,288 (5.4)	59,092 (93.4)
Sex^*
Men	580,933 (49.7)	215,555 (37.1)
Women	587,687 (50.3)	257,049 (43.7)
IMD decile^, ^*
1 (most affluent)	167,558 (14.0)	62,032 (37.0)
2	129,704 (11.0)	51,504 (39.7)
3	128,234 (11.0)	51,794 (40.4)
4	109,986 (9.4)	45,681 (41.5)
5	127,816 (11.0)	53,601 (41.9)
6	104,158 (8.9)	44,279 (42.5)
7	108,782 (9.3)	44,097 (40.5)
8	103,501 (8.9)	43,102 (41.6)
9	100,577 (8.6)	40,019 (39.8)
10 (most deprived)	88,304 (7.6)	36,495 (41.3)

Open in a new tab

*There were statistically significant differences in the proportion of people with and without multimorbidity (P < 0.001) within each variable (row) group using χ² tests.

**IMD.

IMD, Index of Multiple Deprivation.

Of the 80 conditions examined, 6 conditions were present in more than 5% of the whole population: hypertension was the most prevalent (9.1%), followed by depression (8.7%), asthma (7.6%), upper gastro-intestinal (GI) tract acid conditions (7.6%), anxiety (6.7%), and osteoarthritis (5.7%). Nine conditions were present in less than 0.1% of the population (Table 2).

Table 2. Prevalence of individual conditions.

Prevalence rank	Long-term condition	Population count (%) N = 1,168,620	Prevalence rank	Long-term condition	Population count (%) N = 1,168,620
1	Hypertension	212,520 (18.2)	41	Paroxysmal tachycardias	8,747 (0.7)
2	Depression	201,991 (17.3)	42	Obstructive and reflux uropathy	8,501 (0.7)
3	Asthma	177,301 (15.2)	43	Polymyalgia rheumatica	8,446 (0.7)
4	Upper GI tract acid conditions	176,518 (15.1)	44	Intellectual disability	8,004 (0.7)
5	Anxiety	156,762 (13.4)	45	Secondary malignancy	7,332 (0.6)
6	Osteoarthritis	132,799 (11.4)	46	Schizophrenia	7,097 (0.6)
7	Primary malignancy	74,917 (6.4)	47	Haematological malignancy	7,028 (0.6)
8	Type 2 diabetes mellitus	61,671 (5.3)	48	Autism	6,714 (0.6)
9	Chronic kidney disease	59,812 (5.1)	49	Visual impairment and blindness	5,942 (0.5)
10	Coronary heart disease	58,585 (5.0)	50	Obsessive compulsive disorder	5,553 (0.5)
11	Thyroid disease	57,289 (4.9)	51	Bronchiectasis	5,123 (0.4)
12	Erectile dysfunction	53,198 (4.5)	52	Bipolar affective disorder	4,572 (0.4)
13	Diverticular disease	46,410 (4.0)	53	Type 1 diabetes mellitus	4,439 (0.4)
14	Urinary incontinence	45,681 (3.9)	54	Coeliac disease	4,188 (0.4)
15	Psoriasis	41,871 (3.6)	55	Eating disorders	3,766 (0.3)
16	Benign prostatic hyperplasia	33,656 (2.7)	56	Tubulointerstitial nephropathy	3,511 (0.3)
17	Gout	33,483 (2.9)	57	Cardiomyopathy	3,399 (0.3)
18	Atrial fibrillation	33,098 (2.8)	58	Parkinson’s disease	2,938 (0.2)
19	Alcohol misuse	31,841 (2.7)	59	Multiple sclerosis	2,932 (0.2)
20	Chronic obstructive pulmonary disease	31,654 (2.7)	60	Hyperparathyroidism	2,316 (0.2)
21	Osteoporosis	30,284 (2.6)	61	Ankylosing spondylosis	2,214 (0.2)
22	Stroke and transient ischaemic attack	27,925 (2.4)	62	Cerebral palsy	2,051 (0.2)
23	Urolithiasis	25,551 (2.2)	63	Lupus erythematosus	1,752 (0.1)
24	Peripheral neuropathy	24,065 (2.1)	64	Pulmonary fibrosis	1,712 (0.1)
25	Hearing loss	22,535 (1.9)	65	Primary thrombocytopaenia	1,707 (0.1)
26	Venous thromboembolic disease	21,058 (1.8)	66	Giant cell arteritis	1,662 (0.1)
27	Epilepsy	19,139 (1.6)	67	Paralysis	1,625 (0.1)
28	Heart valve disease	18,172 (1.5)	68	Primary pulmonary hypertension	1,210 (0.1)
29	Heart failure	17,879 (1.5)	69	Sjogren syndrome	1,184 (0.1)
30	Substance misuse	17,433 (1.5)	70	Sick sinus syndrome	1,097 (0.1)
31	Inflammatory arthritis	15,853 (1.4)	71	Thalassaemia	1,088 (0.1)
32	Endometriosis	14,567 (1.2)	72	Human immunodeficiency virus	960 (0.1)
33	Sleep apnoea	12,853 (1.1)	73	Diabetes mellitus other or not specified	933 (0.1)
34	Raynaud’s disease	12,312 (1.0)	74	Aplastic anaemia	820 (0.1)
35	Neuropathic bladder	11,507 (1.0)	75	Cystic fibrosis	559 (0.05)
36	Dementia	11,374 (1.0)	76	Scleroderma	442 (0.04)
37	Inflammatory bowel disease	10,949 (0.9)	77	Myasthenia gravis	419 (0.04)
38	Liver disease	10,628 (0.9)	78	Sickle cell disease	415 (0.04)
39	Peripheral arterial disease	10,427 (0.9)	79	Addison’s disease	394 (0.03)
40	Heart block and bundle branch block	10,231 (0.9)	80	Motor neurone disease	231 (0.02)

Open in a new tab

GI, gastro-intestinal.

There was marked variability in multimorbidity prevalence depending on the number of conditions considered in the count. Using all 80 conditions, multimorbidity prevalence was 40.5% (95% CI [40.4, 40.6] p < 0.001). When considering only the 2 conditions most prevalent in the whole population in the count, multimorbidity was present in 4.6% (95% CI [4.6, 4.6] p < 0.001) (Fig 1, S3 Table). When adding more conditions to the count (the most prevalent remaining conditions first), there was a steep increase in estimated multimorbidity prevalence, rising to 29.5% (95% CI [29.5, 29.6%] p < 0.001) when considering 10 conditions in the count. Following this, a more gradual increase in estimated prevalence was seen as more conditions were added to the count: 35.2% (95% CI [35.1, 35.3] p < 0.001) considering 20 conditions, and 37.4% (95% CI [37.3, 37.5] p < 0.001) considering 30 conditions. There was only 0.7 percentage point absolute difference in prevalence between considering 50 conditions, 39.8% (95% CI [39.7, 39.9] p < 0.001), and all 80 conditions (40.5%). In the whole population, the predefined ceiling where adding additional conditions had little impact on prevalence was reached at 52 conditions (i.e., estimated prevalence for 52 conditions versus 80 conditions RR >0.99).

Fig 1 — The black line represents multimorbidity prevalence calculated when considering different numbers of conditions in the count ranging from 2 to all 80 conditions, where conditions were added in order of most to least prevalent (e.g., at 2 conditions this is multimorbidity prevalence considering the most common 2 conditions). Percentage prevalence of multimorbidity when 10, 20, 30, 40, 50, 60, 70, and 80 conditions were considered is marked at empty black circles above the black line. The number of conditions at which RR was >0.99 of multimorbidity prevalence of having the same multimorbidity prevalence when all 80 conditions were considered (ceiling effect) was reached is marked with an orange dot (at 52 conditions). Black dots represent multimorbidity prevalence when considering conditions included in existing condition-lists and are annotated with the condition-list name, prevalence, and number of conditions considered. RR, relative risk.

Multimorbidity prevalence varied widely between the 9 different condition-lists, varying from 11.1% (95% CI [11.0, 11.2] p < 0.001) using the Ho always [10] list, to 36.4% (95% CI [36.3, 36.5] p < 0.001) using the Ho always + usually [10] list (Fig 1, Table 3). Three condition-lists (Ho always + usually [10], Barnett [16], and Fortin [9]) had prevalence close to that estimated by including the same number of the most common conditions in the number of conditions analysis (represented by proximity of these points to the black line in Fig 1). These lists also had highest RR of the multimorbidity prevalence calculated when considering all 80 conditions in the count: Ho always + usually [10] RR 0.90 (95% CI [0.89, 0.90] p < 0.001), Barnett and colleagues [16] RR 0.84 (95% CI [0.83, 0.84] p < 0.001), and Fortin and colleagues [9] RR 0.82 (95% CI [0.82, 0.82] p < 0.001) (Table 2). The remaining 5 condition-lists, however, had prevalence considerably below that estimated by including the same number of most common conditions with prevalence RR 0.27 (95% CI [0.27, 0.27] p < 0.001) and RR 0.27 (95% CI [0.27, 0.27] p < 0.001) respectively using the Ho always [10] and Charlson [18] condition-lists.

Table 3. Multimorbidity prevalence using existing condition-lists and RR of multimorbidity when considering all 80 conditions (ceiling effect where adding additional conditions had little impact on prevalence).

Condition-list	No. of conditions in condition-list	Multimorbidity prevalence No. (%)	RR (95% CI)
All 80 conditions (reference)	80	473,533 (40.5)	1.0
Ho always + usually [10]	55	425,413 (36.4)	0.90 (0.89–0.90)
Barnett [16]	41	397,009 (34.0)	0.84 (0.83–0.84)
N’Goran [11]	29	219,098 (18.7)	0.46 (0.46–0.46)
Elixhauser [19]	28	268,261 (23.0)	0.57 (0.56–0.57)
Fortin [9]	27	389,286 (33.3)	0.82 (0.82–0.82)
Ho always [10]	27	129,698 (11.1)	0.27 (0.27–0.27)
Charlson [18]	21	135,166 (11.6)	0.28 (0.28–0.29)
Diederichs [6]	18	224,001 (19.2)	0.47 (0.47–0.47)
Salive [17]	14	222,859 (19.1)	0.47 (0.47–0.47)

Open in a new tab

RR, relative risk.

The initial gradient of increase in multimorbidity prevalence seen as conditions were added to the count was steepest in the oldest age groups, followed by flattening of the curve as more conditions were considered (Fig 2). In 0–9- and 10–19-year-olds, there was a more gradual increase in prevalence, because rarer conditions contribute to a higher proportion of multimorbidity in children and young people. The influence of adding additional numbers of conditions to the count on estimated prevalence plateaued at a lower number of conditions considered in older people. In people aged 80 years and over, the predefined ceiling (prevalence compared to 80 conditions RR >0.99) was reached at 29 conditions, compared to 71 conditions in those aged 0 to 9 years (Fig 2). In IMD-stratified analysis, there was a clear social gradient of multimorbidity prevalence irrespective of the number of conditions included, with the more deprived having higher prevalence than the less deprived (Fig 3). The predefined ceiling was reached at a lower number of conditions in the most deprived IMD decile (49 conditions) compared to the least deprived (54 conditions) (Fig 3). Sensitivity analysis using raw (unstandardised rates) had less clear social gradient (reflecting that the most deprived are on average younger than the affluent), and no clear pattern in the predefined ceiling across IMD deciles (S1 Fig).

Fig 2 — Labelled coloured lines represent multimorbidity prevalence calculated when considering different numbers of conditions in the count ranging from 2 to all 80 conditions stratified into age groups. Black dots represent the number of conditions at which RR >0.99 of multimorbidity prevalence of having the same multimorbidity prevalence when all 80 conditions were considered (ceiling effect): 0–9 years at 71 conditions, 10–19 years at 67 conditions, 20–29 conditions at 57 conditions, 30–39 years at 57 conditions, 40–49 years at 56 conditions, 50–59 years at 50 conditions, 60–69 years at 44 conditions, 70–79 years at 37 conditions, 80+ years at 29 conditions. RR, relative risk.

Fig 3 — Labelled coloured lines represent multimorbidity prevalence calculated when considering different numbers of conditions in the count ranging from 2 to all 80 conditions stratified into IMD deciles where IMD 1 is least and IMD 10 is most deprived. Black dots represent the number of conditions at which RR >0.99 of multimorbidity prevalence of having the same multimorbidity prevalence when all 80 conditions were considered (ceiling effect): IMD 10 at 49 conditions, IMD 9 at 50 conditions, IMD 8 at 50 conditions, IMD 7 at 51 conditions, IMD 6 at 51 conditions, IMD 5 at 53 conditions, IMD 4 at 52 conditions, IMD 3 at 53 conditions, IMD 2 at 53 conditions, and IMD 1 at 54 conditions. Direct age standardisation where the whole study cohort was the standard population was applied (see S1 Fig for unstandardised rates). IMD, Index of Multiple Deprivation; RR, relative risk; SEP, socioeconomic position.

Multimorbidity prevalence was higher in women and girls at every level of number of conditions in the direct age-standardised analysis, and the predefined ceiling was reached at a higher number of conditions in women and girls than in men and boys (Fig 4). In the sensitivity analysis using unstandardised rates, there was a larger gap in multimorbidity prevalence between sexes, reflecting that women are on average older, and the predefined ceiling was reached at a similar number of conditions to the direct age-standardised analysis (S2 Fig).

Fig 4 — Labelled coloured lines represent multimorbidity prevalence calculated when considering different numbers of conditions in the count ranging from 2 to all 80 conditions stratified by sex. Black dots represent the number of conditions at which RR >0.99 of multimorbidity prevalence of having the same multimorbidity prevalence when all 80 conditions were considered (ceiling effect): women and girls at 54 conditions and men and boys at 50 conditions. Direct age standardisation where the whole study cohort was the standard population was applied (see S2 Fig for unstandardised rates). RR, relative risk.

In both age and deprivation stratified analyses, fewer conditions were required to reach RR 0.99 for the groups with highest prevalence. However, a different pattern was seen in men (who had lower multimorbidity prevalence) where the ceiling was reached at 50 conditions, compared to 54 in women (Fig 4).

The age distribution of multimorbidity prevalence was not uniform across the 9 condition-lists. Across all ages, multimorbidity prevalence using the Ho always + usually [10] condition-list was closest to prevalence when considering all 80 conditions (Fig 5). The Fortin [9], Barnett [16], and Elixhauser [19] condition-lists had lower prevalence than Ho always + usually [10] but followed a similar upward trajectory from youngest to oldest. Salive and colleagues [17] and Diederichs and colleagues [6] had low prevalence in younger age groups, but multimorbidity prevalence increased steeply from age 50 to 59 years and older onwards. Ho always [10] and Charlson [18] had markedly lower prevalence rates than other condition-lists across all age groups.

Fig 5 — Labelled coloured lines represent multimorbidity prevalence calculated for each age group when considering conditions in each condition-list.

Discussion

This study found very large differences in estimated multimorbidity prevalence from varying the number and selection of conditions considered in the count, and in younger people, the more affluent, and women, including additional relatively rare conditions had larger impact on estimated multimorbidity prevalence. Multimorbidity prevalence differed considerably by varying the number of conditions, ranging from 4.6% to 40.5%, and selection of conditions considered, ranging from 11.1% to 36.4% using 9 previously published lists of conditions [6,9,10,16–19]. Counting multimorbidity prevalence using the 9 existing condition-lists resulted in lower estimated prevalence than when considering the same number of the most common conditions, although the extent of this varied: Ho always + usually [10], Fortin [9], and Barnett [16] had the best performance.

Consistent with the findings of this study, there is a wide range in the number and selection of conditions considered in the current multimorbidity literature [4], and in estimates of multimorbidity prevalence [7]. A systematic review of 566 multimorbidity studies by Ho and colleagues [4] found that the number of conditions considered by existing studies ranged from 2 to 285 (median 17, IQR 11 to 23), and very little uniformity in terms of the selection of conditions was found across studies. Only 8 conditions (diabetes, stroke, cancer, chronic obstructive pulmonary disease, hypertension, coronary heart disease, chronic kidney disease, and heart failure) were considered in at least half of the studies, and a quarter of studies did not consider any mental health condition. Simard and colleagues [23] reviewed existing literature to examine how studies used, developed, and validated methods for measuring multimorbidity. They found heterogeneity in the grouping of conditions, validation processes, number of ICD-10 code digits used to define included conditions, and use of additional data sources. Diederichs and colleagues [6] recognised the need to establish a standardised instrument to measure multimorbidity and recommended a minimum set of 11 conditions to include (cancer, diabetes mellitus, depression, hypertension, myocardial infarction, chronic ischemic heart disease, heart arrhythmias, heart insufficiency, stroke, COPD, and arthritis). These conditions were selected based on high prevalence and a severe impact on affected individuals in terms of impairment of function and high need for management, from a population of people aged over 64 years old in Germany. A recent systematic review and meta-analysis of 193 studies examining multimorbidity prevalence [7] did not directly compare prevalence when considering different condition-lists, however, did find that prevalence was significantly higher in studies considering a larger number of conditions in the count: studies considering 44 or more conditions had higher pooled multimorbidity prevalence (87.6%) than studies considering fewer than 9 conditions (30.1%).

Strengths of this study include comprehensive analysis of multimorbidity prevalence estimates in a large population dataset derived from primary care electronic health records. Analysis systematically examined multimorbidity prevalence in the same population for different numbers of conditions considered in the count and using condition-lists recommended or used in previous studies. However, a limitation is that we did not necessarily replicate how previous studies ascertained the presence of conditions, but instead defined the presence of each condition using published UK code lists. This improves comparability within this study but highlights that further variability in prevalence estimates will happen because of variation in how each condition is ascertained (i.e., variation in exactly which codes or prescriptions are used, or restrictions on how recent a diagnosis must be). There is heterogeneity in which conditions are included between existing lists of conditions, and therefore, decisions were made about how to standardise conditions to the full list of 80 conditions. For example, the Barnett (2012) [16] condition-list used time-limited diagnoses that were not replicated in this study in order to make condition ascertainment consistent across the condition-lists examined. Using electronic health records to ascertain the prevalence of conditions can be associated with under ascertainment because the absence of a record does not necessarily mean absence of the condition, and more severe disease is likely to be overrepresented in medical records [24]. The data were from 2015, however, prevalence of the commonest condition hypertension was unchanged between 1990 and 2019 in a pooled analysis of worldwide population studies (32% of women in 2019 versus 32% in 1990) [25], and this study identified similar prevalence rates of depression as ascertained by an Office of National Statistics survey from 2021 [26].

Deciding which conditions to include in multimorbidity research is complicated, including in the extent to which conditions should be aggregated (e.g., coronary heart disease) or considered separately (e.g., angina, myocardial infarction). Ideally, researchers would use a standardised list to improve research comparability and reproducibility, but this is not always feasible due to varying data availability and varying prevalence of disease in different settings. An alternative method is to use an “open condition-list,” as used by Fortin and colleagues [8] in a Canadian study where methodology was not constrained to a specified number of conditions considered in the count to calculate multimorbidity prevalence, but considered all conditions present in a patient’s medical records in the multimorbidity count. The number of conditions in study participants was highly variable and resulted in large differences in multimorbidity prevalence, particularly for younger people [7]. The method of data collection involved manual review of medical records, and therefore, although this analysis provides additional richness, it would be challenging to scale this approach and apply it to larger populations.

Even where researchers agree on which conditions to measure, there is an additional source of variation introduced by heterogeneity in methods chosen to measure and ascertain those conditions in data. Based on this research, if the purpose of the study is to estimate prevalence then estimates will be relatively stable providing the 50 most common conditions are considered; however, this threshold requires examination in other datasets and settings. Although some tailoring to local context and purpose will often be necessary, comparability and reproducibility would be improved by choices always starting with a core list of conditions. Researchers should therefore consider using the Delphi consensus derived Ho always + usually list [10], or for measuring prevalence the Barnett [16], or Fortin [9] condition-lists.

There are several areas where further research is needed. First, this study examined relationships between the number and selection of conditions and multimorbidity prevalence in the UK. However, similar studies in low- and middle-income countries are needed, where prevalence of individual conditions will be different. Second, condition ascertainment in routine data is based on lists of clinical codes (and sometimes prescribing or laboratory data) [27]. However, there can be large variations in the clinical codes used to define the same condition in different studies [28]. Therefore, further exploration of the impact of variation in which codes or prescriptions are used to define conditions is needed. Applying the condition codes from validated open-access published code lists, such as the HDR-UK Phenotype Library [15] or other similar sources [29] will also improve comparability and reproducibility. Third, it is important to examine whether and how much the number and selection of conditions considered in counts alter observed associations with important clinical outcomes such as functional status, unplanned hospital use, and death.

The key implication of this study is that the choice of conditions to consider when estimating multimorbidity prevalence has a large impact on the results, with additional variation in impact between older versus younger people particularly. The comparability and reproducibility of multimorbidity research would be improved by researchers including recommended core conditions wherever possible [10], with explicitly justified variation for study context and purpose.

Supporting information

S1 Table. List of chronic conditions examined and codes used to define those morbidities.

(DOCX)

Click here for additional data file.^{(22.5KB, docx)}

S2 Table. Conditions in each condition-list as implemented in this study, and as stated in published condition-list.

(DOCX)

Click here for additional data file.^{(51.9KB, docx)}

S3 Table. The RECORD statement.

Checklist of items, extended from the STROBE statement, that should be reported in observational studies using routinely collected health data.

(DOCX)

Click here for additional data file.^{(22.9KB, docx)}

S1 Information Panel. Measuring multimorbidity in research: a Delphi consensus study.

Two recommended condition-lists defined by a modified Delphi panel study [10] were used. This study developed international consensus on the measurement of multimorbidity in research and was funded by HDR-UK. Data were collected in 3 rounds of online questionnaires, including 25 public panel and 150 professional panel members. Public members had an interest in, or personal experience of, multimorbidity. Professional participants were clinicians, policy makers, and researchers interested in, or involved in, multimorbidity work. Two sets of questions were developed separately for the public and professional panels. Questions in subsequent rounds were informed by results from the previous questionnaire. Participants were asked to answer open and closed questions, where open questions were subsequently triangulated by subsequent closed questions. Consensus was reached for 24 conditions to always include in multimorbidity measures, and 35 conditions to usually include unless a good reason not to, and these lists have been examined in our study of multimorbidity prevalence. In the study, we calculated multimorbidity prevalence using the always include list (Ho always) and both condition-lists together (Ho always + usually).

(EPS)

Click here for additional data file.^{(3.6MB, eps)}

S1 Fig. Socioeconomic position stratified multimorbidity prevalence according to number of conditions considered without direct age standardisation, reporting the ceiling effect where adding additional conditions had little impact on prevalence.

Labelled coloured lines represent multimorbidity prevalence calculated when considering different numbers of conditions in the count ranging from 2 to all 80 conditions stratified into IMD deciles where IMD 1 is least and IMD 10 is most deprived. Black dots represent the number of conditions at which RR >0.99 of multimorbidity prevalence of having the same multimorbidity prevalence when all 80 conditions were considered (ceiling effect): IMD 10 at 51 conditions, IMD 9 at 53 conditions, IMD 8 at 51 conditions, IMD 7 at 51 conditions, IMD 6 at 51 conditions, IMD 5 at 52 conditions, IMD 4 at 51 conditions, IMD 3 at 52 conditions, IMD 2 at 52 conditions, and IMD 1 at 54 conditions.

(TIFF)

Click here for additional data file.^{(82.4MB, tiff)}

S2 Fig. Sex stratified multimorbidity prevalence according to number of conditions considered without direct age standardisation, reporting the ceiling effect where adding additional conditions had little impact on prevalence.

Labelled coloured lines represent multimorbidity prevalence calculated when considering different numbers of conditions in the count ranging from 2 to all 80 conditions stratified by sex. Black dots represent the number of conditions at which RR >0.99 of multimorbidity prevalence of having the same multimorbidity prevalence when all 80 conditions were considered (ceiling effect): women and girls at 54 conditions and men and boys at 51 conditions.

(TIFF)

Click here for additional data file.^{(82.4MB, tiff)}

Acknowledgments

Dr. Niall Anderson for providing statistical consultancy.

Abbreviations

CPRD: Clinical Practice Research Datalink
GI: gastro-intestinal
HDR-UK: Health Data Research UK
IMD: Index of Multiple Deprivation
IQR: interquartile range
RR: relative risk
SEP: socioeconomic position

Data Availability

All relevant data are within the manuscript or available via GitHub repository at https://github.com/macraec. Raw data cannot be made publicly available to protect data security. The third-party contact for access to raw data is CPRD’s Research Data Governance (RDG) Process at rdg@cprd.com.

Funding Statement

This work was supported by the Chief Scientist Office (HIPS/18/30) to BG, SWM, DA, EJ, DM, NHS Education for Scotland Academic Fellowship for CM, and Medical Research Council MR/W000253/1 fellowship for CM. This study/project is funded by the National Institute for Health Research (NIHR) Artificial Intelligence and Multimorbidity: Clustering in Individuals, Space and Clinical Context (AIM-CISC) grant NIHR202639. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.Pefoyo AJK, Bronskill SE, Gruneir A, Calzavara A, Thavorn K, Petrosyan Y, et al. The increasing burden and complexity of multimorbidity. BMC Public Health. 2015;15(1):415. doi: 10.1186/s12889-015-1733-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Violan C, Foguet-Boreu Q, Flores-Mateo G, Salisbury C, Blom J, Freitag M, et al. Prevalence, determinants and patterns of multimorbidity in primary care: a systematic review of observational studies. PLoS ONE. 2014;9(7):e102149–e. doi: 10.1371/journal.pone.0102149 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Multimorbidity: a priority for global health research. The Academy of Medical Sciences. Available from: https://acmedsci.ac.uk/policy/policy-projects/multimorbidity [accessed 2022 Apr 18].
4.Ho IS-S, Azcoaga-Lorenzo A, Akbari A, Black C, Davies J, Hodgins P, et al. Examining variation in the measurement of multimorbidity in research: a systematic review of 566 studies. Lancet Public Health. 2021;6(8):e587–e97. doi: 10.1016/S2468-2667(21)00107-9 [DOI] [PubMed] [Google Scholar]
5.Johnston MC, Crilly M, Black C, Prescott GJ, Mercer SW. Defining and measuring multimorbidity: a systematic review of systematic reviews. Eur J Public Health. 2019;29(1):182–189. doi: 10.1093/eurpub/cky098 [DOI] [PubMed] [Google Scholar]
6.Diederichs C, Berger K, Bartels DB. The measurement of multiple chronic diseases—a systematic review on existing multimorbidity indices. J Gerontol A Biol Sci Med Sci. 2011;66(3):301–11. doi: 10.1093/gerona/glq208 [DOI] [PubMed] [Google Scholar]
7.Ho IS-S, Azcoaga-Lorenzo A, Akbari A, Davies J, Hodgins P, Khunti K, et al. Variation in the estimated prevalence of multimorbidity: systematic review and meta-analysis of 193 international studies. BMJ Open. 2022;12(4):e057017–e. doi: 10.1136/bmjopen-2021-057017 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Fortin M, Hudon C, Haggerty J, van den Akker M, Almirall J. Prevalence estimates of multimorbidity: a comparative study of two sources. BMC Health Serv Res. 2010;10(1):111. doi: 10.1186/1472-6963-10-111 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Fortin M, Almirall J, Nicholson K. Development of a Research Tool to Document Self-Reported Chronic Conditions in Primary Care. J Comorb. 2017;7(1):117–23. doi: 10.15256/joc.2017.7.122 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Ho ISS, Azoaga-Lorenzo A, Akbari A, Davies J, Khunti K, Kadam U, et al. Measuring multimorbidity in research: a Delphi consensus study. BMJ Medicine. July 2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Herrett E, Gallagher AM, Bhaskaran K, Forbes H, Mathur R, van Staa T, et al. Data Resource Profile: Clinical Practice Research Datalink (CPRD). Int J Epidemiol. 2015;44(3):827–36. doi: 10.1093/ije/dyv098 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.The English Indices of Deprivation 2019. Ministry of Housing, Communities and Local Goverment. Available from: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/853811/IoD2019_FAQ_v4.pdf [accessed 2022 Mar 27].
13.Lewis JD, Bilker WB, Weinstein RB, Strom BL. The relationship between time since registration and measured incidence rates in the General Practice Research Database. Pharmacoepidemiol Drug Saf. 2005;14(7):443–51. doi: 10.1002/pds.1115 [DOI] [PubMed] [Google Scholar]
14.Kuan V, Denaxas S, Gonzalez-Izquierdo A, Direk K, Bhatti O, Husain S, et al. A chronological map of 308 physical and mental health conditions from 4 million individuals in the English National Health Service. Lancet Digit Health. 2019;1(2):e63–e77. doi: 10.1016/S2589-7500(19)30012-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.N’Goran AA, Blaser J, Deruaz-Luyet A, Senn N, Frey P, Haller DM, et al. From chronic conditions to relevance in multimorbidity: A four-step study in family medicine. Fam Pract. 2016;33(4):439–44. doi: 10.1093/fampra/cmw030 [DOI] [PubMed] [Google Scholar]
16.Barnett K, Mercer SW, Norbury M, Watt G, Wyke S, Guthrie B. Epidemiology of multimorbidity and implications for health care, research, and medical education: a cross-sectional study. Lancet. 2012;380(9836):37–43. doi: 10.1016/S0140-6736(12)60240-2 [DOI] [PubMed] [Google Scholar]
17.Salive ME. Multimorbidity in older adults. Epidemiol Rev. 2013;35(1):75–83. doi: 10.1093/epirev/mxs009 [DOI] [PubMed] [Google Scholar]
18.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis. 1987;40(5):373–83. doi: 10.1016/0021-9681(87)90171-8 [DOI] [PubMed] [Google Scholar]
19.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity Measures for Use with Administrative Data. Medical Care. 1998;36(1):8–27. doi: 10.1097/00005650-199801000-00004 [DOI] [PubMed] [Google Scholar]
20.RDocumentation. ageadjust.direct: Age standardization by direct method. Available from: https://www.rdocumentation.org/packages/epitools/versions/0.09/topics/ageadjust.direct [accessed 2022 Jun 14].
21.Benchimol EI, Smeeth L, Guttmann A, Harron K, Moher D, Petersen I, et al. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med. 2015;12(10):e1001885–e. doi: 10.1371/journal.pmed.1001885 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.R-Core-Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. Vienna, Austria. Available from: http://www.R-project.org [accessed 2021 Mar 27].
23.Simard M, Rahme E, Calfat AC, Sirois C. Multimorbidity measures from health administrative data using ICD system codes: A systematic review. Pharmacoepidemiol Drug Saf. 2022;31(1):1–12. doi: 10.1002/pds.5368 [DOI] [PubMed] [Google Scholar]
24.Bagley SC, Altman RB. Computing disease incidence, prevalence and comorbidity from electronic medical records. J Biomed Inform. 2016;63:108–11. doi: 10.1016/j.jbi.2016.08.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Paciorek CJ, Singleton RK, Ikeda N, Laxmaiah A, Widyahening IS, Abarca-Gómez L, et al. Worldwide trends in hypertension prevalence and progress in treatment and control from 1990 to 2019: a pooled analysis of 1201 population-representative studies with 104 million participants. Lancet. 2021;398(10304):957–80. doi: 10.1016/S0140-6736(21)01330-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Office for National Statistics. Coronavirus and depression in adults in Great Britain. Available from: https://www.ons.gov.uk/peoplepopulationandcommunity/wellbeing/datasets/coronavirusanddepressioninadultsingreatbritain [accessed 2022 Dec 19].
27.MacRae C, Whittaker H, Mukherjee M, Daines L, Morgan A, Iwundu C, et al. Deriving a Standardised Recommended Respiratory Disease Codelist Repository for Future Research. Pragmat Obs Res. 2022;13:1. doi: 10.2147/POR.S353400 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Quint JK, Müllerova H, DiSantostefano RL, Forbes H, Eaton S, Hurst JR, et al. Validation of chronic obstructive pulmonary disease recording in the Clinical Practice Research Datalink (CPRD-GOLD). BMJ Open. 2014;4(7):e005540–e. doi: 10.1136/bmjopen-2014-005540 [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Tonelli M, Wiebe N, Fortin M, Guthrie B, Hemmelgarn BR, James MT, et al. Methods for identifying 30 chronic conditions: application to administrative data. BMC Med Inform Decis Mak. 2015;15(1):31. doi: 10.1186/s12911-015-0155-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

PLoS Med. doi: 10.1371/journal.pmed.1004208.r001

Decision Letter 0

Philippa Dodd

14 Oct 2022

Dear Dr MacRae,

Thank you for submitting your manuscript entitled "Measuring multimorbidity: impact of varying the number and selection of conditions on estimated multimorbidity prevalence in a large primary care population dataset" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff as well as by an academic editor with relevant expertise and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by Oct 18 2022 11:59PM.

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Philippa Dodd, MBBS MRCP PhD

Senior Editor

PLOS Medicine

PLoS Med. doi: 10.1371/journal.pmed.1004208.r002

Decision Letter 1

Philippa Dodd

14 Dec 2022

Dear Dr. MacRae,

Thank you very much for submitting your manuscript "Measuring multimorbidity: impact of varying the number and selection of conditions on estimated multimorbidity prevalence in a large primary care population dataset" (PMEDICINE-D-22-03375R1) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Jan 04 2023 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Philippa Dodd, MBBS MRCP PhD

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

GENERAL

Please respond to all editor and reviewer requests detailed below, in full.

Please number the lines in the manuscript starting at 1 at “Abstract” and in continuous sequence thereafter.

We agree with the reviewers (please see below) that the study’s primary goal should be better emphasized/identified and its implications. Further details and clarifications are required throughout, which we agree with. Please revise accordingly.

** We agree that the novelty is a bit lost among the formal data reporting. Please revise in mind of the potential impact of the study outcomes at the patient/policy level **

COMMENTS FROM THE ACADEMIC EDITOR

I reviewed the article and the reviews and I agree with the decision for major revision

ABSTRACT

Please structure your abstract using the PLOS Medicine headings (Background, Methods and Findings, Conclusions).

Please combine the Methods and Findings sections into one section, “Methods and findings”.

Abstract Methods and Findings:

Please ensure that all numbers presented in the abstract are present and identical to numbers presented in the main manuscript text.

Please provide further details on the population (and setting) that the study investigates

Please provide further details as to how conditions were “selected” when considering multimorbidity prevalence (diagnostic coded registers?)

Please clearly define the main outcome measures.

Please quantify the main results with p-values as well as 95% CIs, if not for the purpose of transparent data reporting please provide the reasons why

Please include any important dependent variables that are adjusted for in the analyses.

Please include a summary of adverse events if these were assessed in the study.

In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

Abstract Conclusions:

Please replace the sub-heading “interpretation” with “conclusions”

Please address the study implications without overreaching what can be concluded from the data; the phrase "In this study, we observed ..." may be useful.

Please interpret the study based on the results presented in the abstract, emphasizing what is new and what the wider impact of the study outcomes are, without overstating your conclusions.

Please avoid vague statements such as "these results have major implications for policy/clinical care". Mention only specific implications substantiated by the results.

Please avoid assertions of primacy ("We report for the first time....")

AUTHOR SUMMARY

At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

INTRODUCTION

Please see reviewer comments – we agree that limitations to the age of the data-set should be discussed, please do so later in the discussion

METHODS and RESULTS

Page 4, data sources: please define SEP at first use in this section

Did your study have a prospective protocol or analysis plan? Please state this (either way) early in the Methods section.

a) If a prospective analysis plan (from your funding proposal, IRB or other ethics committee submission, study protocol, or other planning document written before analyzing the data) was used in designing the study, please include the relevant prospectively written document with your revised manuscript as a Supporting Information file to be published alongside your study, and cite it in the Methods section. A legend for this file should be included at the end of your manuscript.

b) If no such document exists, please make sure that the Methods section transparently describes when analyses were planned, and when/why any data-driven changes to analyses took place.

c) In either case, changes in the analysis-- including those made in response to peer review comments-- should be identified as such in the Methods section of the paper, with rationale.

For all observational studies, we ask authors to indicate the following clearly in the manuscript text:

(1) the specific hypotheses you intended to test,

(2) the analytical methods by which you planned to test them,

(3) the analyses you actually performed, and

(4) when reported analyses differ from those that were planned, transparent explanations for differences that affect the reliability of the study's results. If a reported analysis was performed based on an interesting but unanticipated pattern in the data, please be clear that the analysis was data-driven.

Please quantify the main results with p-values as well as 95% CIs, if not for the purpose of transparent data reporting please provide the reasons why

When a p value is given, please specify the statistical test used to determine it.

Please remove role of the funding source from the end of the methods section and include only in the manuscript submission form.

TABLES and FIGURES

We agree the reviewer that the figures rather undersell the data. The titles are rather inaccessible and difficult to understand and don’t explain the figure contents clearly for the reader. Please revise throughout, including the supporting files.

To make figures more accessible to those with colorblindness please consider avoiding the use of red/green

Please ensure that each figure/table (including those in supporting files) has an appropriate caption

Please define all abbreviations (for example: IMD, RR) in the figure/table captions

Please indicate whether analyses are adjusted and if so, please state the caption inwhich factors are adjusted for and report unadjusted analyses for comparison

DISCUSSION

Please present and organize the Discussion as follows: a short, clear summary of the article's findings; what the study adds to existing research and where and why the results may differ from previous research; strengths and limitations of the study; implications and next steps for research, clinical practice, and/or public policy; one-paragraph conclusion. Please remove the sub-heading conclusion.

Please expand your discussion to better focus on the study aim and implication of the outcomes, without overstating – see also reviewer 2 comments

Please remove all “declaration” statements from the end of the discussion and include only in the submission form. The section “Author contributions” may remain under its own heading.

REFERENCES

Please use the "Vancouver" style for reference formatting, and see our website for other reference guidelines here: https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references

For in-text reference call outs citations should be placed in square brackets and preceding punctuation like so [1,2,3,4] or [1-4,6]. Please amend throughout.

In the bibliography please list up to but no more than 6 author names followed by et al (where more than 6 authors contribute to the study).

Journal name abbreviations should be those found in the National Center for Biotechnology Information (NCBI) databases.

Comments from the reviewers:

Reviewer #1: Statistical review

This paper uses data from a large set of patients in England to investigate how different definitions of multimorbidity affect prevalence estimates. The authors also present this stratified by socioeconomic level, sex and age.

Generally the paper presents an important message for multimorbidity research. The statistical methods used are straightforward, as just prevalence estimates and confidence intervals are needed. I had very few comments, which are presented below:

1. Statistical methods: "To examine the effect of the number of conditions considered in the count, multimorbidity prevalence was calculated when considering the most common two conditions, followed by the most common three conditions, the most common four conditions, etc, for every number up to considering all 80 conditions in the count (Information Panel). To do this, conditions were ordered from most to least prevalent (Supplementary Table 1) and added in turn to each new count" Although this procedure would give the highest multimorbidity prevalence in the case that conditions are independent from one another, it's possible it might not do so if common conditions are correlated. E.g. if considering conditions 1, 2 and 3 in descending prevalence, the prevalence of '1 and 2' might be less than '1 and 3' if conditions 1 and 2 co-occur in a lot of patients. This is just a picky point really and I don't think it makes sense to change what the authors did in any way!

2. Statistical methods: can some more intuition be provided about the direct standardisation to the age structure of the population?

3. Page 8: "29.5% (95%CI 29.5-29.9%)" - is this correct? The lower boundary and the point estimate are the same.

4. Discussion: are the prevalence estimates generally an underestimate due to potential under-diagnosis of some conditions?

James Wason

Reviewer #2: Thank you for the opportunity to review this work. In their manuscript, the authors explore how various definitions of multimorbidity influence its estimated prevalence. As the authors acknowledge, the growing attention to multimorbidity - and its highly varied operationalization - has contributed to methodological variation and biases in the literature. Overall, this manuscript covers an important aspect of multimorbidity research.

I have a few comments for the authors to consider. Overall, though, I think the paper would benefit from identifying the primary goal and really tightening up the methods, results, and discussion around this. It was, at times, a bit hard to follow. I am not advocating for removing any of the current analyses, but rather making sure they are presented in a way that easily conveys the overarching theme and story while highlighting the important and novel aspects of this work.

Major:

1. Introduction: the authors state "it is common for authors not to explain the procedure used to decide which conditions they consider in the count" - I don't think this is necessarily true. Instead, there's often a 'recycling' of previously published lists. This could be reframed to highlight that the heterogeneity the authors are exploring is not due to authors not presenting information.

2. The authors acknowledge in the introduction that the relationship between multimorbidity and time is important to note given its growing prevalence. However, the data the authors analyze is now nearly 7 years old. While I doubt it would make any meaningful change in the results, the authors should consider acknowledging the datedness of these data and whether they believe this influences their findings at all.

3. Would the authors consider making Supplemental Table 1 part of the main text? I think the list of included conditions is the crux of this study. I know it's visually unappealing, but I think could be important.

4. Could the authors, in the methods, more clearly state that they included every condition mention in any of those studies? It seems to me that is what the authors did, but I think this needs to be made explicitly clear.

5. The approach to use a relative risk and evaluate the appropriate 'cutoff' given those 80 conditions is an interesting approach but could be better setup in the introduction and the methods. It's a bit confusing, right now, the various comparisons that are being made. Indeed, I'd encourage the authors to consider an approach where they explicitly and succinctly lay out all their analyses. For example, "We conducted a suite of comparisons including: ____."

6. The relative risk analysis seems to be the most novel part of this studied but is not as prominently featured. I would encourage the authors to restructure their manuscript focusing on this if they also believe it's the most novel. Right now, some of the most interesting findings feel a bit buried in the analysis.

a. Similarly, it's unclear what the intermediary analyses contribute. For example, "When considering only the two conditions…" as presented in the results. While I understand this was a step along the way, I'm not sure that any study would ever pick only two conditions to estimate the prevalence of multimorbidity.

b. Additionally, I think the first sentence of the discussion doesn't highlight the most important findings. Of course, when it's 2 versus 80 there is a large difference. There are several other findings that are better to highlight - for example the sentence beginning "Second, multimorbidity…"

Minor:

1. The authors could remove the date from the abstract methods.

2. The abstract findings are hard to follow without the additional context presented in the full manuscript. I would encourage the authors to try to restructure their findings to make it more interpretable.

3. Could the authors instead spell out "November 30, 2015" since DD/MM/YY is not standard globally.

4. Please mention the funders in the methods.

5. Please define "IMD" in Table 1.

Reviewer #3: Dear editor,

Thank you for allowing me to read such an interesting manuscript.

I believe that the manuscript was well written and that the research was presented in a way that was easy to follow and understand. I think the statistics were adequately explained and their meaning was easy to understand.

I have only one question for the authors: what were the criteria for choosing some condition lists like the ones from Diederichs, Fortin, Ho, Barnett, Salive, but not others that are well know such as O'Halloran and N'Goran (particularly in primary care settings)?

- O'Halloran J, Miller GC, Britt H. Defining chronic conditions for primary care with ICPC-2. Fam Pract. 2004 Aug;21(4):381-6. doi: 10.1093/fampra/cmh407. PMID: 15249526.

- N'Goran AA, Blaser J, Deruaz-Luyet A, Senn N, Frey P, Haller DM, Tandjung R, Zeller A, Burnand B, Herzig L. From chronic conditions to relevance in multimorbidity: a four-step study in family medicine. Fam Pract. 2016 Aug;33(4):439-44. doi: 10.1093/fampra/cmw030. Epub 2016 May 6. PMID: 27154549.

Best regards.

Any attachments provided with reviews can be seen via the following link:

[LINK]

PLoS Med. 2023 Apr 4;20(4):e1004208. doi: 10.1371/journal.pmed.1004208.r003

Author response to Decision Letter 1

24 Jan 2023

Attachment

Submitted filename: PLOS_Medicine_reviewer_response_Jan_2023.docx

Click here for additional data file.^{(77.1KB, docx)}

PLoS Med. doi: 10.1371/journal.pmed.1004208.r004

Decision Letter 2

Philippa Dodd

8 Feb 2023

Dear Dr. MacRae,

Thank you very much for re-submitting your manuscript "Measuring multimorbidity: impact of varying the number and selection of conditions on estimated multimorbidity prevalence in a large primary care population dataset" (PMEDICINE-D-22-03375R2) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by 3 reviewers. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Feb 15 2023 11:59PM.

Sincerely,

Philippa Dodd, MBBS MRCP PhD

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

GENERAL

Thank you for your detailed and considerate responses to previous editor and reviewer requests. Please see below for further minor revisions.

DATA AVAILABILITY STATEMENT

Thank you for making your data available.

A point for clarification - in one section of the manuscript submission form you state that some restrictions will apply to your data availability but subsequently you go on to state that all relevant data are included in the manuscript and supporting files. I have detailed our data availability statement below for your information, please clarify/revise/expand your statement as necessary.

PLOS Medicine requires that the de-identified data underlying the specific results in a published article be made available, without restrictions on access, in a public repository or as Supporting Information at the time of article publication, provided it is legal and ethical to do so. Please see the policy at

http://journals.plos.org/plosmedicine/s/data-availability

and FAQs at

http://journals.plos.org/plosmedicine/s/data-availability#loc-faqs-for-data-policy

PLOS defines the “minimal data set” to consist of the data set used to reach the conclusions drawn in the manuscript with related metadata and methods, and any additional data required to replicate the reported study findings in their entirety. Authors do not need to submit their entire data set, or the raw data collected during an investigation. Please submit the following data:

The values behind the means, standard deviations and other measures reported;

The values used to build graphs;

The points extracted from images for analysis.

For each data source used in your study:

a) If the data are freely or publicly available, note this and state the location of the data: within the paper, in Supporting Information files, or in a public repository (include the DOI or accession number).

b) If the data are owned by a third party but freely available upon request, please note this and state the owner of the data set and contact information for data requests (web or email address). Note that a study author cannot be the contact person for the data.

c) If the data are not freely available, please describe briefly the ethical, legal, or contractual restriction that prevents you from sharing it. Please also include an appropriate contact (web or email address) for inquiries (again, this cannot be a study author).

TITLE

Suggest the following to align better with PLOS Medicine’s preferred style,

The impact of varying the number and selection of conditions on estimated multimorbidity prevalence: a cross sectional study using a large, primary-care population dataset

STATISITICAL REPORTING

Thank you for including p values. Please ensure that throughout the manuscript including in the tables, figures and supplementary files p is reported as p <0.001 (not .001 or .0001) and where higher as p=0.002 (not .002), for example.

Throughout, suggest reporting statistical information as follows: X% (95%CI [X,Y] p <0.001)

- Note the use of square parentheses around upper and lower confidence limits

- Note the use of a comma instead of a hyphen (as these can be confused with negative values) to separate upper

and lower limits

- Note the reporting of p values as detailed above

ABSTRACT

Line 30: suggest “We conducted….” Or something similar

Line 32: sentence beginning “Outcome…” is rather long and could be more accessible, please revise

Line 35: please ensure that HDR[-UK] is defined for the reader

Lines 38 - 46: the word is “prevalence is mentioned a number of times, might it be helpful to remind the reader that you are reporting multimorbidity prevalence at some point, we leave it to your discretion bt it may be worth considering.

Line 42: is the word “prevalence” required twice here or is this a typo? Please clarify/revise

Line 54 - 55: we suggest avoiding naming the lists here as the reader is yet to learn what all the lists are. See comments below also. Suggest, “These findings imply that there is a need for a standardised approach to defining multimorbidity.” Then go on to state (concisely) how your study helps to solve this problem.

AUTHOR SUMMARY

Line 64: “in half of studies…” suggest make into a separate bullet point

Line 67: suggest “2” instead of “two to 285 (median 17).”

Line 68: perhaps one line stating why? i.e. comparability of study outcomes, generalisability etc

Line 72: suggest “and the impact on multimorbidity prevalence” perhaps?

Line 73: It would be helpful for the reader to elaborate here – what were the differences, briefly

Line 75: this won’t mean anything to the reader as it hasn’t been introduced anywhere in the text apart the abstract but without adequate context (see above for a similar point) suggest either removing/modifying this statement or revising the summary (and abstract) accordingly such that it’s clear to the reader what this is/means. For example, you could expand bullet point 1 of this sub-section to include (some examples) of the lists you used to derive counts and then go on to state which of these were most reliable, as we understand things.

Line 83: suggest removing this statement as it is best placed as part of a discussion

** The early part of your introduction, description of study design in your methods section and, the early part of your discussion summarise things very well and may be helpful with the revisions to your author summary **

REFERENCES

Please ensure that punctuation follows, not precedes, in-text reference call-outs. For example, line 94, “…systems worldwide.[1]” should read, “…systems worldwide [1]. Please check and amend throughout

SOCIAL MEDIA

To help us extend the reach of your research, please provide any Twitter handle(s) that would be appropriate to tag, including your own, your coauthors’, your institution, funder, or lab. Please detail any handles you wish to be included when we tweet this paper, in the manuscript submission form when you re-submit the manuscript.

Comments from Reviewers:

Reviewer #1: Thank you to the authors for addressing my previous comments well. I have no further issues to raise.

Reviewer #2: The authors have thoughtfully addressed the points raised. Thank you.

Reviewer #3: The authors have responded to the remaining comments made by me.

Thank you and best wishes.

Any attachments provided with reviews can be seen via the following link:

[LINK]

PLoS Med. doi: 10.1371/journal.pmed.1004208.r005

Decision Letter 3

Philippa Dodd

20 Feb 2023

Dear Dr. MacRae,

Thank you very much for re-submitting your manuscript "The impact of varying the number and selection of conditions on estimated multimorbidity prevalence: a cross-sectional study using a large, primary care population dataset" (PMEDICINE-D-22-03375R3) for review by PLOS Medicine.

I have discussed the paper with my colleagues and I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

[LINK]

We expect to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Feb 23 2023 11:59PM.

Sincerely,

Philippa Dodd, MBBS MRCP PhD

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

GENERAL

Thank you for your responses to previous editor requests.

Your author summary needs further revision before we can publish your manuscript. Please see below

AUTHOR SUMMARY

* What did the researchers do and find – this section should be no more than 4 bullet points.

Without reading other parts of the manuscript, it’s difficult to understand what you did. Further some points, to me seem to be illogically ordered. Please see specific points below and revise accordingly.

Line 74 and line 78: suggest combining (and shortening) these points into one point with additional details about what the “lists” are, for example, “We examined nine published condition lists used to define and measure co-morbidity, multi-morbidity and its prevalence. We combined different numbers and types of conditions (detailed in these lists?) to determine how multimorbidity prevalence changed with varying classifications/definitions… All conditions were counted in the same way….” or something similar

The statement currently at line 76 should then follow the above

Line 84: should precede that at line 81

** What do these findings mean – this section should explain the implications of your study findings only.

Line 90: The first point here justifies your methodology and seems inappropriately placed. The sentence is also rather long. Please remove this. If you think this point is important to make, then it should be made in the “what did the researchers do and find” section (see above).

Line 96: When reading the full summary, it suggests that using lists which report “highest” multimorbidity prevalence is advisable. Please (concisely) elaborate as to why this is the case – do these yield more consistent results (?), for example

Any attachments provided with reviews can be seen via the following link:

[LINK]

PLoS Med. doi: 10.1371/journal.pmed.1004208.r006

Decision Letter 4

Philippa Dodd

27 Feb 2023

Dear Dr MacRae,

On behalf of my colleagues and the Academic Editor, Professor Aaron Kesselheim, I am pleased to inform you that we have agreed to publish your manuscript "The impact of varying the number and selection of conditions on estimated multimorbidity prevalence: a cross-sectional study using a large, primary care population dataset" (PMEDICINE-D-22-03375R4) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

Thank you again for submitting to PLOS Medicine, it has been a pleasure handling your manuscript. We look forward to publishing your paper.

Best wishes,

Pippa

Philippa Dodd, MBBS MRCP PhD

PLOS Medicine

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 Table. List of chronic conditions examined and codes used to define those morbidities.

(DOCX)

Click here for additional data file.^{(22.5KB, docx)}

S2 Table. Conditions in each condition-list as implemented in this study, and as stated in published condition-list.

(DOCX)

Click here for additional data file.^{(51.9KB, docx)}

S3 Table. The RECORD statement.

Checklist of items, extended from the STROBE statement, that should be reported in observational studies using routinely collected health data.

(DOCX)

Click here for additional data file.^{(22.9KB, docx)}

S1 Information Panel. Measuring multimorbidity in research: a Delphi consensus study.

(EPS)

Click here for additional data file.^{(3.6MB, eps)}

(TIFF)

Click here for additional data file.^{(82.4MB, tiff)}

Labelled coloured lines represent multimorbidity prevalence calculated when considering different numbers of conditions in the count ranging from 2 to all 80 conditions stratified by sex. Black dots represent the number of conditions at which RR >0.99 of multimorbidity prevalence of having the same multimorbidity prevalence when all 80 conditions were considered (ceiling effect): women and girls at 54 conditions and men and boys at 51 conditions.

(TIFF)

Click here for additional data file.^{(82.4MB, tiff)}

Attachment

Submitted filename: PLOS_Medicine_reviewer_response_Jan_2023.docx

Click here for additional data file.^{(77.1KB, docx)}

Data Availability Statement

[pmed.1004208.ref001] 1.Pefoyo AJK, Bronskill SE, Gruneir A, Calzavara A, Thavorn K, Petrosyan Y, et al. The increasing burden and complexity of multimorbidity. BMC Public Health. 2015;15(1):415. doi: 10.1186/s12889-015-1733-2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref002] 2.Violan C, Foguet-Boreu Q, Flores-Mateo G, Salisbury C, Blom J, Freitag M, et al. Prevalence, determinants and patterns of multimorbidity in primary care: a systematic review of observational studies. PLoS ONE. 2014;9(7):e102149–e. doi: 10.1371/journal.pone.0102149 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref003] 3.Multimorbidity: a priority for global health research. The Academy of Medical Sciences. Available from: https://acmedsci.ac.uk/policy/policy-projects/multimorbidity [accessed 2022 Apr 18].

[pmed.1004208.ref004] 4.Ho IS-S, Azcoaga-Lorenzo A, Akbari A, Black C, Davies J, Hodgins P, et al. Examining variation in the measurement of multimorbidity in research: a systematic review of 566 studies. Lancet Public Health. 2021;6(8):e587–e97. doi: 10.1016/S2468-2667(21)00107-9 [DOI] [PubMed] [Google Scholar]

[pmed.1004208.ref005] 5.Johnston MC, Crilly M, Black C, Prescott GJ, Mercer SW. Defining and measuring multimorbidity: a systematic review of systematic reviews. Eur J Public Health. 2019;29(1):182–189. doi: 10.1093/eurpub/cky098 [DOI] [PubMed] [Google Scholar]

[pmed.1004208.ref006] 6.Diederichs C, Berger K, Bartels DB. The measurement of multiple chronic diseases—a systematic review on existing multimorbidity indices. J Gerontol A Biol Sci Med Sci. 2011;66(3):301–11. doi: 10.1093/gerona/glq208 [DOI] [PubMed] [Google Scholar]

[pmed.1004208.ref007] 7.Ho IS-S, Azcoaga-Lorenzo A, Akbari A, Davies J, Hodgins P, Khunti K, et al. Variation in the estimated prevalence of multimorbidity: systematic review and meta-analysis of 193 international studies. BMJ Open. 2022;12(4):e057017–e. doi: 10.1136/bmjopen-2021-057017 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref008] 8.Fortin M, Hudon C, Haggerty J, van den Akker M, Almirall J. Prevalence estimates of multimorbidity: a comparative study of two sources. BMC Health Serv Res. 2010;10(1):111. doi: 10.1186/1472-6963-10-111 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref009] 9.Fortin M, Almirall J, Nicholson K. Development of a Research Tool to Document Self-Reported Chronic Conditions in Primary Care. J Comorb. 2017;7(1):117–23. doi: 10.15256/joc.2017.7.122 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref010] 10.Ho ISS, Azoaga-Lorenzo A, Akbari A, Davies J, Khunti K, Kadam U, et al. Measuring multimorbidity in research: a Delphi consensus study. BMJ Medicine. July 2022. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref011] 11.Herrett E, Gallagher AM, Bhaskaran K, Forbes H, Mathur R, van Staa T, et al. Data Resource Profile: Clinical Practice Research Datalink (CPRD). Int J Epidemiol. 2015;44(3):827–36. doi: 10.1093/ije/dyv098 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref012] 12.The English Indices of Deprivation 2019. Ministry of Housing, Communities and Local Goverment. Available from: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/853811/IoD2019_FAQ_v4.pdf [accessed 2022 Mar 27].

[pmed.1004208.ref013] 13.Lewis JD, Bilker WB, Weinstein RB, Strom BL. The relationship between time since registration and measured incidence rates in the General Practice Research Database. Pharmacoepidemiol Drug Saf. 2005;14(7):443–51. doi: 10.1002/pds.1115 [DOI] [PubMed] [Google Scholar]

[pmed.1004208.ref014] 14.Kuan V, Denaxas S, Gonzalez-Izquierdo A, Direk K, Bhatti O, Husain S, et al. A chronological map of 308 physical and mental health conditions from 4 million individuals in the English National Health Service. Lancet Digit Health. 2019;1(2):e63–e77. doi: 10.1016/S2589-7500(19)30012-3 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref015] 15.N’Goran AA, Blaser J, Deruaz-Luyet A, Senn N, Frey P, Haller DM, et al. From chronic conditions to relevance in multimorbidity: A four-step study in family medicine. Fam Pract. 2016;33(4):439–44. doi: 10.1093/fampra/cmw030 [DOI] [PubMed] [Google Scholar]

[pmed.1004208.ref016] 16.Barnett K, Mercer SW, Norbury M, Watt G, Wyke S, Guthrie B. Epidemiology of multimorbidity and implications for health care, research, and medical education: a cross-sectional study. Lancet. 2012;380(9836):37–43. doi: 10.1016/S0140-6736(12)60240-2 [DOI] [PubMed] [Google Scholar]

[pmed.1004208.ref017] 17.Salive ME. Multimorbidity in older adults. Epidemiol Rev. 2013;35(1):75–83. doi: 10.1093/epirev/mxs009 [DOI] [PubMed] [Google Scholar]

[pmed.1004208.ref018] 18.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis. 1987;40(5):373–83. doi: 10.1016/0021-9681(87)90171-8 [DOI] [PubMed] [Google Scholar]

[pmed.1004208.ref019] 19.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity Measures for Use with Administrative Data. Medical Care. 1998;36(1):8–27. doi: 10.1097/00005650-199801000-00004 [DOI] [PubMed] [Google Scholar]

[pmed.1004208.ref020] 20.RDocumentation. ageadjust.direct: Age standardization by direct method. Available from: https://www.rdocumentation.org/packages/epitools/versions/0.09/topics/ageadjust.direct [accessed 2022 Jun 14].

[pmed.1004208.ref021] 21.Benchimol EI, Smeeth L, Guttmann A, Harron K, Moher D, Petersen I, et al. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med. 2015;12(10):e1001885–e. doi: 10.1371/journal.pmed.1001885 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref022] 22.R-Core-Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. Vienna, Austria. Available from: http://www.R-project.org [accessed 2021 Mar 27].

[pmed.1004208.ref023] 23.Simard M, Rahme E, Calfat AC, Sirois C. Multimorbidity measures from health administrative data using ICD system codes: A systematic review. Pharmacoepidemiol Drug Saf. 2022;31(1):1–12. doi: 10.1002/pds.5368 [DOI] [PubMed] [Google Scholar]

[pmed.1004208.ref024] 24.Bagley SC, Altman RB. Computing disease incidence, prevalence and comorbidity from electronic medical records. J Biomed Inform. 2016;63:108–11. doi: 10.1016/j.jbi.2016.08.005 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref025] 25.Paciorek CJ, Singleton RK, Ikeda N, Laxmaiah A, Widyahening IS, Abarca-Gómez L, et al. Worldwide trends in hypertension prevalence and progress in treatment and control from 1990 to 2019: a pooled analysis of 1201 population-representative studies with 104 million participants. Lancet. 2021;398(10304):957–80. doi: 10.1016/S0140-6736(21)01330-1 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref026] 26.Office for National Statistics. Coronavirus and depression in adults in Great Britain. Available from: https://www.ons.gov.uk/peoplepopulationandcommunity/wellbeing/datasets/coronavirusanddepressioninadultsingreatbritain [accessed 2022 Dec 19].

[pmed.1004208.ref027] 27.MacRae C, Whittaker H, Mukherjee M, Daines L, Morgan A, Iwundu C, et al. Deriving a Standardised Recommended Respiratory Disease Codelist Repository for Future Research. Pragmat Obs Res. 2022;13:1. doi: 10.2147/POR.S353400 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref028] 28.Quint JK, Müllerova H, DiSantostefano RL, Forbes H, Eaton S, Hurst JR, et al. Validation of chronic obstructive pulmonary disease recording in the Clinical Practice Research Datalink (CPRD-GOLD). BMJ Open. 2014;4(7):e005540–e. doi: 10.1136/bmjopen-2014-005540 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1004208.ref029] 29.Tonelli M, Wiebe N, Fortin M, Guthrie B, Hemmelgarn BR, James MT, et al. Methods for identifying 30 chronic conditions: application to administrative data. BMC Med Inform Decis Mak. 2015;15(1):31. doi: 10.1186/s12911-015-0155-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

The impact of varying the number and selection of conditions on estimated multimorbidity prevalence: A cross-sectional study using a large, primary care population dataset

Clare MacRae

Megan McMinn

Stewart W Mercer

David Henderson

David A McAllister

Iris Ho

Emily Jefferson

Daniel R Morales

Jane Lyons

Ronan A Lyons

Chris Dibben

Bruce Guthrie

Roles

Abstract

Background

Methods and findings

Conclusions

Author summary

Why was this study done?

What did the researchers do and find?

What do these findings mean?

Introduction

Methods

Study design

Data sources

Study participants

Definition of variables

Statistical analysis

Results

Table 1. Population characteristics.

Table 2. Prevalence of individual conditions.

Fig 1. Multimorbidity prevalence according to number of conditions, the ceiling effect where adding additional conditions had little impact on prevalence, and selection of conditions using existing condition-lists.

Table 3. Multimorbidity prevalence using existing condition-lists and RR of multimorbidity when considering all 80 conditions (ceiling effect where adding additional conditions had little impact on prevalence).

Fig 2. Age-stratified multimorbidity prevalence according to number of conditions considered, reporting the ceiling effect where adding additional conditions had little impact on prevalence.

Fig 3. SEP-stratified multimorbidity prevalence according to number of conditions considered following direct age standardisation, reporting the ceiling effect where adding additional conditions had little impact on prevalence.

Fig 4. Sex-stratified multimorbidity prevalence according to number of conditions considered following direct age standardisation, reporting the ceiling effect where adding additional conditions had little impact on prevalence.

Fig 5. Multimorbidity prevalence by age considering all 80 conditions and according to existing condition-lists.

Discussion

Supporting information

Acknowledgments

Abbreviations

Data Availability

Funding Statement

References

Decision Letter 0

Philippa Dodd

Roles

Decision Letter 1

Philippa Dodd

Roles

Author response to Decision Letter 1

Decision Letter 2

Philippa Dodd

Roles

Decision Letter 3

Philippa Dodd

Roles

Decision Letter 4

Philippa Dodd

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases