Relationship satisfaction change among UK service personnel: Associations with socio-demographics, military factors, and mental health

ABSTRACT

The unique demands of military life can adversely impact romantic relationships; however, research has mainly focused on these adverse outcomes at one-time point, overlooking changes over time or potential positive outcomes. Using a subsample of 3,845 male and female military personnel and veterans from a large UK dataset, this study examined positive and negative changes in relationship satisfaction between two-time points (2007–2009; 2014–2016). Most participants reported no change in their relationship satisfaction, suggesting stability − 8% reported a positive change and 10% a negative change. Positive change was associated with being in a long-term relationship, alcohol misuse remission, and persistent alcohol misuse. Negative change was associated with the onset of mental health problems (probable PTSD, CMD, or alcohol misuse) and having children under 18. Some factors, like increasing age, childhood family relationship adversity, and mental health problems, were associated with both positive and negative changes in relationship satisfaction. This study highlights the complexity of factors associated with relationship satisfaction among military personnel and veterans, with some experiencing positive changes, as well as negative changes over time.

What is the public significance of this article?— This study offers a nuanced understanding of changes in self-reported relationship satisfaction among serving and ex-serving personnel. While most service personnel reported no change (82%) in their relationship satisfaction, a small minority reported experiencing either a positive (8%) or negative (10%) change in their relationship satisfaction over time. Factors such as being in a long-term relationship and alcohol misuse remission were associated with positive changes, while mental health difficulties and having children under 18 were associated with negative changes in satisfaction. A key finding from this study is that some factors, such as age, childhood family adversity, and mental health challenges, were associated with both positive and negative changes in relationship satisfaction. These findings highlight the complexity of factors associated with relationship satisfaction that should be considered when supporting service personnel and encourage researchers to look at both negative and positive outcomes where possible.

Introduction

The military is a unique occupational context that can present various challenges to service personnel and their families. The demands of military life, including frequent relocations, operational deployments, and schedule inflexibility, have been associated with adverse effects on the maintenance and stability of romantic relationships of service personnel (Burrell et al., 2006; Keeling et al., 2015; Knobloch-Fedders et al., 2020). In both the United States (US) and the United Kingdom (UK), studies estimate that between 10% and 30% of service personnel report experiencing relationship dissatisfaction, marital distress, or other relationship difficulties (Caska-Wallace et al., 2019; Keeling et al., 2015; Meis et al., 2017). Furthermore, military occupational demands have been found to negatively influence the well-being of civilian partners (Caska-Wallace et al., 2019; Gribble et al., 2019), potentially leading to diminished morale, decreased work efficiency, and challenges in service personnel retention (Keeling et al., 2015).

Several factors intrinsic to military service have been shown to influence relationship satisfaction, including lower rank (Anderson et al., 2011; Karney & Crown, 2007), deployment and combat exposure (Newby et al., 2005; Schumm et al., 2000). Prolonged separation due to deployment may contribute to emotional distress and increased conflict, especially during the reintegration phase (Allen et al., 2010). In this regard, US and UK-based research further highlights the negative impact of the separation from one’s partner during extended deployments on relationship satisfaction (Keeling et al., 2015; Riviere et al., 2012; Rona et al., 2009). The transition from military to civilian life can also be a significant life stressor, contributing to increased relationship challenges during the readjustment period among ex-serving personnel (Elnitsky et al., 2017).

Within the military, gender dynamics have also been suggested to impact relationship satisfaction. Women in Service often face distinct stressors, such as balancing military and family responsibilities, which can influence relationship satisfaction differently compared to their male counterparts – this is further complicated among dual serving couples (Southwell & MacDermid Wadsworth, 2016). However, UK-based research suggests that gender differences in relationship satisfaction among service personnel are not consistently found, calling for further investigation (Keeling et al., 2015),

Consistent with the general population (Wilson & Stuchbury, 2010), socio-demographic factors, such as younger age, cohabiting rather than marriage, and lower educational and socioeconomic backgrounds have been associated with relationship dissatisfaction within military populations (Keeling et al., 2015; Riviere et al., 2012). Additionally, experiencing adverse childhood experiences, such as exposure to family conflict or abuse, a common pre-enlistment vulnerability among UK service personnel has been suggested to negatively influence intimacy and trust in adulthood, contributing to increased relationship difficulties (Iversen et al., 2007; Keeling et al., 2015).

Mental health challenges, such as posttraumatic stress disorder (PTSD) (Campbell & Renshaw, 2013; Caska-Wallace et al., 2019; Erbes et al., 2012) and depression (Allen et al., 2010; Foran et al., 2013; Knobloch & Theiss, 2011; Riviere et al., 2012) have been consistently associated with relationship dissatisfaction among serving and ex-serving personnel. These conditions have been found to exacerbate communication difficulties, result in emotional withdrawal, and reduce reported intimacy (Foran et al., 2013; Miller et al., 2013). These findings have been replicated among ex-serving, serving, and reserve personnel within US military samples (Allen et al., 2010; Meis et al., 2017), and among the general population (DeCuyper et al., 2018; Taft et al., 2011). Alcohol misuse, particularly prevalent among both UK and US active service personnel, has similarly been linked to reduced relationship satisfaction and negative relationship dynamics (Mattiko et al., 2011; Meis et al., 2010; Riviere et al., 2012).

Despite these challenges, military life can also foster resilience and strengthen relationships among some couples. Shared experiences, such as navigating deployment stressors and overcoming adversity together, may enhance emotional bonds and mutual understanding (Knobloch-Fedders et al., 2020). Moreover, structured support systems, such as military family programs and initiatives such as the Families OverComing Under Stress FOCUS) program, can help mitigate stressors and promote relationship satisfaction by enhancing communication and resilience (Burrell et al. 2006); Saltzman et al. 2011).

To date, much of the literature has examined relationship satisfaction among military personnel and veterans using a cross-sectional approach. While this provides useful estimates of prevalence and possible risk and protective factors, there are limitations to our understanding of relationship satisfaction dynamics among military populations over time. Moreover, existing research has tended to adopt a deficit approach, focusing on associations with adverse outcomes rather than exploring the possible benefits military life may offer some couples. The present study uses an exploratory approach to address these gaps by (1) investigating both positive and negative changes in relationship satisfaction among serving and ex-serving UK service personnel across two-time points; and (2) identifying associations between changes in relationship satisfaction and socio-demographic, military and family factors, and mental health outcomes.

Methods

Data collection

Data comes from the King’s Centre for Military Health Research’s (KCMHR) Health and Well-being cohort study, established in 2003 to examine the health of UK military personnel deployed to Iraq (Hotopf et al., 2006). To date, three phases of data collection of serving and ex-serving personnel have occurred. In brief, Phase 1 (2004–2005) compared UK Armed Forces personnel deployed to Iraq in 2003 to non-deployed serving personnel (59% response, n  = 10,272) (Hotopf et al., 2006). Phase 2 (2007–2009) followed up Phase 1 participants and collected data from two additional random samples: service personnel deployed to Afghanistan (n = 1,789), and newly enlisted service personnel (n  = 6,628), (56% response, n  = 9,984) (Fear et al., 2010). In Phase 3 (2014–2016), all participants were followed up (n  = 12,280), and a representative sample of newly enlisted service personnel (n  = 8,581) was invited to take part (44.3% response, n = 8,093) (Stevelink et al., 2018). Ethical approval was granted by the Ministry of Defence Research Ethics Committee (reference: 448/MODREC/13) and the King’s College London Psychiatry, Nursing and Midwifery Research Ethics Subcommittee (Reference: PNM/12/13–169).

Procedure

At all phases, participants completed questionnaires covering: socio-demographics (age, gender, education; marital status); service information (including service branch (Royal Navy/Marines, Army, Royal Air Force), current or final rank (Officer, Non-Commissioned Officers (NCOs), Other (Enlisted, Private-Equivalent)), serving status (serving/left Service), engagement type (Regular/Reserve), deployment to Iraq and Afghanistan (not deployed at p2 or p3,/deployed p2 only, deployed p3 only, deployed at both p2 and p3); role in parent unit (combat/other); current mental and physical health; and lifestyle behaviors. Relationship satisfaction was collected in Phase 2 (p2) and Phase 3 (p3) only.

Participants

All participants who completed both p2 and p3, who reported being in a relationship, and who rated their relationship satisfaction were included (n = 3,845) (Table 1). Service personnel not in a relationship (n = 3,484 at p2, n = 1,519 at p3) and those in a relationship but who did not respond to the question on relationship satisfaction at either phase (n = 3,491 at p2, n = 1,551 at p3) were excluded. While both serving and ex-serving were included, henceforth all participants will be referred to as “service personnel.”

Table 1.

Sociodemographic and military profile of participant at phase 2 and phase 3.

	Phase 2		Phase 3
	n	%	n	%
Sociodemographic characteristics
Gender
Male	3,465	90.1	3,465	90.1
Female	380	9.9	380	9.9
Age (in years)
<40 (20–39)	2,072	56.2	1,154	32.4
40–49	1,319	32.9	1,530	40.2
50+	454	10.9	1,161	27.4
Marital Status
Married	2,884	74.5	3,354	86.4
Cohabiting	522	14.5	376	10.4
In a long-term relationship	439	11.0	115	2.2
Children <18 years of age
No	1,525	39.0	1,578	40.7
Yes	2,306	61.0	2,204	59.3
Education
None – O levels/GCSEs	1,279	39.0	959	27.8
A levels	1,303	36.9	1,174	32.7
Degree or above	1,100	24.1	1,703	39.6
Military Characteristics
Service Branch
Royal Navy/Royal Marines	594	17.4	594	17.4
Army	2,370	60.9	2,370	60.9
Royal Air Force	881	21.7	881	21.7
Rank
Officers	1,211	22.9	1,311	24.7
NCOs	2,200	63.6	2,269	66.1
Others	434	13.5	265	9.2
Serving Status
Serving	2,766	66.6	1,735	35.8
Left Service	1,053	33.4	2,107	64.2
Engagement Type
Regulars	3,115	89.5	3,115	89.5
Reserves	730	10.5	730	10.5
Deployment to Iraq/Afghanistan
Yes	2,512	53.3	2,729	60.4
No	1,333	46.7	1,116	38.6
Combat Role in Parent Unit
Combat support/service support	3,033	78.4	2,886	74.3
Combat role	788	21.6	956	25.7
Mental Health Caseness
Probable PTSD (PCL-5/C)
No (score <50/38)	3,727	97.1	3,658	95.1
Yes (score ≥50/38)	98	2.9	155	4.9
Probable CMD (GHQ-12)
No (score <4)	3,162	82.2	3,098	81.1
Yes (score ≥4)	658	17.8	726	18.9
Alcohol misuse (AUDIT)
No (score <16)	3,499	90.9	3,502	90.8
Yes (score ≥16)	318	9.1	331	9.2

Missing = 3–163 NB: Participants at p2 and p3 are the same individuals.

NCOs = Non-Commissioned Officers.

Measures

Changes in relationship satisfaction

Relationship satisfaction was determined by asking participants at p2 and p3, “How satisfied are you with your marriage/relationship?.” In line with Keeling et al. (2015), responses were collapsed into three categories: satisfied (extremely satisfied/satisfied), neither satisfied/dissatisfied, and dissatisfied (dissatisfied/extremely dissatisfied).

Change in relationship satisfaction was determined by comparing responses from p2 to p3 (see Table 2). Participants were classified as “No Change” if the assessment remained consistent (e.g., satisfied to satisfied), “Positive Change” if their satisfaction improved (i.e., dissatisfied to neither satisfied/dissatisfied; dissatisfied to satisfied; neither satisfied/dissatisfied to satisfied), and “Negative Change” if their satisfaction decreased (i.e., satisfied to dissatisfied; satisfied to neither satisfied/dissatisfied; neither satisfied nor dissatisfied to dissatisfied).

Table 2.

Associations between change in relationship satisfaction and socio-demographic characteristics.

Change in Relationship Satisfaction
				Unadjusted ORa		Adjusted ORb
Socio–demographic characteristics	No Change n = 3,137 (82.0%)	Positive Change n = 315(8.0%)	Negative Change n = 393(10.0%)	Positive Change OR [95% CI]	Negative Change OR [95% CI]	Positive Change OR [95% CI]	Negative Change OR [95% CI]
Gender
Male	2,829 (82.0)	284 (8.0)	352 (10.0)	1.00	1.00	–	–
Female	308 (82.2)	31 (7.7)	41 (10.1)	0.95 [0.60, 1.50]	1.01 [0.66, 1.54]	–	–
Age (in years) (p2)
<40	1,653 (80.5)	179 (8.6)	240 (10.9)	1.00	1.00	1.00	1.00
40–49	1,089 (81.5)	112 (7.7)	118 (10.8)	0.88 [0.65, 1.18]	0.71 [0.54, 0.95]*	0.91 [0.67, 1.23]	0.75 [0.56, 0.99]*
50+	395 (84.6)	24 (7.7)	35 (7.7)	0.55 [0.34, 0.90]*	0.49 [0.31, 0.76]**	0.61 [0.37, 1.01]*	0.54 [0.34, 0.84]**
Education (p2)
Degree or above	905 (82.5)	80 (7.4)	115 (10.1)	1.00	1.00
A levels	1,060 (82.4)	118 (8.5)	125 (9.2)	1.14 [0.80, 1.62]	0.91 [0.66, 1.26]	–	–
None – O Levels/GCSEs	1,031 (80.6)	109 (8.4)	139 (11.0)	1.15 [0.81, 1.65]	1.12 [0.82, 1.53]	–	–
CFRA
0–1	2,138 (85.5)	159 (6.2)	225 (8.3)	1.00	1.00	1.00	1.00
2 or more	924 (75.0)	151 (11.6)	162 (13.5)	2.12 [1.60, 2.80]***	1.85 [1.43, 2.40]***	2.53 [1.76, 3.65]**	1.60 [1.18, 2.16]***

Missing 5–141.

^aOR Baseline Category: No Change; Male; < 40 years of age; Degree or above education, Childhood relationship adversity 0–1.

^bAdjusted for: Age at Phase 2, Childhood Family Relationship Adversity (CFRA).

*p ≤ 05; **p ≤ .01; ***p ≤ .001.

Sociodemographic characteristics

Associations between changes in relationship satisfaction and baseline sociodemographic characteristics (age (years); sex (male/female); educational attainment (none/O levels/GCSEs; A levels; Degree or above)) were examined. Childhood family relationship adversity was assessed via an adaptation of the Adverse Childhood Experience Scale (Felitti et al., 1998) composed of 4-positive reversed score items (e.g., “I came from a close family”) and 4-negative items (e.g., “I used to be hit/hurt by a parent or caregiver regularly”) (Iversen et al., 2007). These eight items were calculated as a cumulative measure and categorized as 0–1 or 2 or more adversities.

Military characteristics

Associations between changes in relationship satisfaction and baseline military characteristics (Service branch (Royal Navy/Royal Marines; Army; Royal Air Force), rank (Officers; Non-Commissioned Officers (NCOs); Others), and engagement type (Regular/Reserves)) were examined. Changes in deployment status (not deployed to Iraq or Afghanistan; deployed p2 only; deployed p3 only; deployed both phases), changes in parent unit role (combat support/service support at both phases; combat p2 only; combat p3 only; combat both phases) and changes in serving status (still serving; left Service at p2; left Service at p3) were also explored.

Family circumstances

Associations between changes in relationship satisfaction and family characteristics were examined. This included relationship type at p2 (married; cohabiting; long-term relationship) and parental status at either phase (no children below 18 years of age; children below 18 years of age).

Changes in mental health and alcohol misuse

Changes in mental health and alcohol misuse were determined by categorizing participants as “Not Case” if they did not meet probable caseness at p2 or p3; “Persistent Case” if they were probable cases at both phases; “Remission” if they were probable cases at p2 only; or “New Onset” if they were probable cases at p3 only.

Common mental health disorders (CMD) were assessed using the 12-item General Health Questionnaire (GHQ-12 (Goldberg et al., 1997); scores 0 to 12; caseness ≥ 4; Cronbach’s α p2 = 0.97, p3 = 0.93) and alcohol misuse using the 10-item Alcohol Use Disorders Identification Test (AUDIT (Babor et al., 2001); scores 0 to 40; caseness ≥ 16; Cronbach’s α p2 = 0.93, p3 = 0.77).

Two different measures were used to assess probable post-traumatic stress disorder (PTSD) due to changes in conceptualization between p2 and p3. At p2, PTSD was assessed via the 17-item National Centre for PTSD Checklist (PCL-C (Blanchard et al., 1996); scores 17 to 85; caseness ≥ 50; Cronbach’s α = 0.97). At p3, the 20-item National Centre for PTSD Checklist was used (PCL-5 (Weathers et al., 2013); scores 0–80; caseness ≥ 38; Cronbach’s α = 0.94). These cutoffs were chosen for consistency with prior work among military populations (Bovin et al., 2016; Weathers et al., 1993). Although these measures differ slightly in structure and scoring, they have demonstrated comparability in assessing probable PTSD (LeardMann et al., 2021; Moshier et al., 2019).

Statistical analysis

Response rates were calculated to account for non-response as the inverse probability of responding based on predicting factors (see Stevelink et al., 2018). All analyses were conducted using Stata® (StataCorp, 2019).

The sociodemographic and military characteristics of the sample, change in relationship satisfaction, as well as cross-tabulations between relationship satisfaction change and exploratory variables are reported using weighted percentages and unweighted counts (Table 2). Given limited cases in the sample, mental health comorbidities were not explored.

Multinomial regressions were used to examine associations between change in relationship satisfaction and key factors of interest due to multiple response categories in the outcome. Univariate multinomial regressions were conducted to identify sociodemographic, military, family and mental health factors significantly associated with changes in relationship satisfaction at p < .05 and significant findings were used to construct six multivariable regression models using a stepwise approach. Model 1 adjusted for sociodemographic variables significantly associated with changes in relationship satisfaction (Table 2), model 2 comprised model 1 adjusted for military factors (Table 3), and model 3 adjusted for model 1 and model 2, as well as changes in family circumstances (Table 4). The final three models (Table 5) adjusted for models 1–3 and changes in probable PTSD (Model 4), probable CMD (Model 5), and alcohol misuse (Model 6), respectively. To limit potential interpretation errors, no adjustments for multiple comparisons were implemented (Rothman, 1990).

Table 3.

Associations between change in relationship satisfaction and military characteristics.

Change in Relationship Satisfaction
				Unadjusted OR a			Adjusted ORb
Military characteristics	No Change n = 3,137 (82.0%)	Positive Change n = 315(8.0%)	Negative Change n = 393(10.0%)	Positive Change OR [95% CI]	Negative Change OR [95% CI]		Positive Change OR [95% CI]	Negative Change OR [95% CI]
Service Branch (p2)
RN/RM	491 (82.3%)	50 (8.9%)	53 (6.9%)	1.05 [0.72, 1.55]	0.60 [0.42, 0.87]**		1.16 [0.77, 1.75]	0.67 [0.45, 0.98]*
Army	1,919 (81.0%)	202 (8.1%)	249 (10.9%)	1.00	1.00		1.00	1.00
RAF	727 (83.0%)	63 (7.1%)	91 (9.9%)	0.85 [0.60, 1.21]	0.89 [0.66, 1.20]		0.94 [0.66, 1.34]	1.04 [0.76, 1.43]
Rank (p2)
Officer	1,014 (84.5%)	82 (6.4%)	115 (9.1%)	1.00	1.00		1.00	1.00
NCOs	1,771 (81.0%)	193 (8.5%)	236 (10.5%)	1.38 [1.01, 1.87]*	1.20 [0.91, 1.58]		1.12 [0.81, 1.57]	1.03 [0.76, 1.39]
Other	352 (82.4%)	40 (8.5%)	42 (9.0%)	1.37 [0.87, 2.16]	1.01 [0.65, 1.57]		1.08 [0.63, 1.84]	0.80 [0.48, 1.33]
Engagement Type
Regulars	2,565 (82.3%)	245 (7.8%)	305 (9.9%)	1.00	1.00		–	–
Reserves	572 (79.6%)	70 (10.0%)	88 (10.4%)	1.33 [0.95, 1.85]	1.08 [0.79, 1.47]		–	–
Change in deployment status
Not Deployed	863 (84.2%)	71 (7.1%)	95 (8.7%)	1.00	1.00		1.00	1.00
Deployed (p3)	248 (83.3%)	29 (8.4%)	27 (8.2%)	1.21 [0.73, 1.99]	0.95 [0.58, 1.56]		1.19 [0.68, 2.10]	0.77 [0.46, 1.28]
Deployed (p2)	71 (80.8%)	8 (10.8%)	9 (8.4%)	1.59 [0.64, 3.96]	1.01 [0.42, 2.42]		1.69 [0.67, 4.28]	0.96 [0.40, 2.32]
Deployed (p2, p3)	1,955 (80.2%)	207 (8.5%)	263 (11.3%)	1.26 [0.92, 1.73]	1.37 [1.03, 1.81]*		1.16 [0.79, 1.69]	0.97 [0.71, 1.33]
Change in Combat Role (Parent Unit)
Combat support p2 & p3	2,214 (82.2%)	223 (8.1%)	278 (9.7%)	1.00	1.00		–	–
Combat role p2	121 (78.3%)	14 (9.4%)	17 (12.3%)	1.22 [0.63, 2.36]	1.34 [0.72, 2.50]		–	–
Combat role p3	253 (82.5%)	26 (6.4%)	37 (11.1%)	0.78 [0.47, 1.31]	1.15 [0.72, 1.82]		–	–
Combat role p2 & p3	524 (81.6%)	51 (8.1%)	60 (10.3%)	1.00 [0.68, 1.46]	1.08 [0.76, 1.53]		–	–
Change in Serving Status
Still Serving	1,350 (79.7%)	133 (8.4%)	187 (11.9%)	1.00	1.00		1.00	1.00
Left Service (p2)	852 (85.5%)	75 (7.1%)	76 (7.4%)	0.78 [0.55, 1.01]	0.58 [0.42, 0.81]***		0.92 [0.63, 1.35]	0.69 [0.48, 0.99]*
Left Service (p3)	875 (81.4%)	99 (8.3%)	119 (10.3%)	0.97 [0.70, 1.34]	0.85 [0.63, 1.14]		1.03 [0.74, 1.45]	0.91 [0.67, 1.24]

Missing 1–60.

^aOR Baseline Category: No Change, Army at p2, Officer at p2, Still Serving, Regular, Not Deployed at either phase; Combat support at p2 and p3.

^bAdjusted for: Age at p2, Childhood Family Relationship Adversity, Service Branch, Rank at p2, Change in Serving Status, Change in Deployment Status.

* p≤ 05; ** p ≤ .01; ***p ≤ .001.

Table 4.

Association between change in relationship satisfaction and family circumstances.

Change in Relationship Satisfaction
				Unadjusted OR a		Adjusted ORb
Family circumstances	No Change n = 3,137 (82.0%)	Positive Change n = 315(8.0%)	Negative Change n = 393(10.0%)	Positive Change OR [95% CI]	Negative Change OR [95% CI]	Positive Change OR [95% CI]	Negative Change OR [95% CI]
Relationship Status (p2)
Married	2,381 (82.9)	212 (7.2)	291 (9.8)	1.00	1.00	1.00	1.00
Living with partner	416 (80.7)	50 (8.6)	56 (10.7)	1.22 [0.83, 1.80]	1.12 [0.77, 1.64]	1.24 [0.82, 1.88]	1.21 [0.80, 1.81]
In long-term relationship	340 (77.6)	53 (12.3)	46 (10.2)	1.82 [1.24, 2.65]**	1.11 [0.74, 1.64)	1.78 [1.16, 2.72]**	0.94 [0.61, 1.46]
Children <18 years of age
No	1,311 (85.2)	137 (7.8)	126 (7.0)	1.00	1.00	1.00	1.00
Yes	1,763 (79.9)	173 (8.2)	259 (9.9)	1.13 [0.86, 1.49]	1.81 [1.37, 2.37]***	1.00 [0.71, 1.39]	1.53 [1.09, 2.14]**

Missing 5–63.

^aOR Baseline Category: No Change, Married at p2, No Children < 18 at either phase.

^bAdjusted for: Age at p2, Childhood Family Relationship Adversity, Service Branch, Change in Serving Status, Relationship Status at p2, Children <18 years of age.

*p ≤ 05; ** p ≤ .01; ***p ≤ .001.

Table 5.

Association between change in relationship satisfaction, change in mental health status, and alcohol misuse.

Change in Relationship Satisfaction
				Unadjusted OR a		Adjusted ORb
Change in Mental Health Caseness	No Change n = 3,137 (82.0%)	Positive Change n = 315(8.0%)	Negative Change n = 393(10.0%)	Positive Change OR [95% CI]	Negative Change OR [95% CI]	Positive Change OR [95% CI]	Negative Change OR [95% CI]
Probable PTSD
Not case (< 50/38 p2, p3)	2,975 (83.7)	281 (7.4)	331 (8.9)	1.00	1.00	1.00	1.00c
Persistent case (≥ 50/38 p2, p3)	24 (52.3)	8 (22.7)	12 (25.0)	4.90 [1.97, 12.19]***	4.49 [1.90, 10.63]***	4.40 [1.70, 11.35]**	4.43 [1.80, 10.93]***
Remission (< 50 p3, ≥ 38 p2)	26 (42.0)	18 (38.1)	9 (19.9)	10.24 [5.08, 20.65]***	4.45 [1.85, 10.72]***	10.95 [5.19, 23.11]***	4.71 [1.83, 12.09]***
New onset (≥ 50 p3, <38 p2)	66 (63.9)	6 (5.5)	38 (30.6)	0.97 [0.36, 2.58]	4.51 [2.76, 7.36]***	0.83 [0.28, 2.44]	4.57 [2.75, 7.60]***
Probable CMD
Not Case (< 4 p2, p3)	2,346 (87.4)	158 (5.7)	195 (6.9)	1.00	1.00	1.00	1.00 d
Persistent Case (≥ 4 p2, p3)	168 (58.6)	51 (19.8)	60 (21.6)	5.21 [3.45, 7.89]***	4.67 [3.12, 7.00]***	4.29 [2.76, 6.67]***	4.23 [2.79, 6.43]***
Remission (< 4 p3, >4 p2)	259 (69.8)	74 (17.9)	47 (12.3)	3.95 [2.77, 5.65]***	2.22 [1.48, 3.35]***	3.82 [2.65, 5.52]***	2.00 [1.29, 3.10]**
New Onset (≥ 4 p3, <4 p2)	326 (74.9)	28 (5.1)	88 (20.0)	1.06 [0.63, 1.72]	3.37 [2.42, 4.69]***	0.92 [0.52, 1.65]	3.34 [2.37, 4.69]***
Alcohol misuse
Not case (< 16 p2, p3)	2,742 (83.9)	248 (7.0)	313 (9.1)	1.00	1.00	1.00	1.00 e
Persistent Case (≥ 16 p2, p3)	98 (68.1)	23 (18.4)	21 (13.4)	3.23 [1.81, 5.76]***	1.82 [1.02, 3.25]*	2.65 [1.42, 4.95]**	1.51 [0.82; 2.78]
Remission (< 16 p3, ≥16 p2)	129 (74.5)	25 (14.6)	29 (10.9)	2.35 [1.39, 3.97]**	1.35 [0.77, 2.38]	2.04 [1.20, 3.47]**	0.93 [0.48, 1.80]
New Onset (≥ 16 p3, <16 p2)	133 (67.5)	17 (9.8)	37 (22.7)	1.73 [0.97, 3.09]	3.10 [1.99, 4.85]***	1.49 [0.80, 2.78]	2.89 [1.79, 4.67]***

Missing 2–46.

^aOR Baseline Category: No Change, Not probable PTSD case at p2 or p3; Not probable CMD case at p2 or p3; No alcohol misuse at p2 or p3.

^bAll adjusted for: Age at p2, Childhood Family Relationship Adversity, Service Branch, Serving Status, Relationship Status at p2, Children <18 years of age.

^cAlso adjusted for Change in PTSD.

^dAlso adjusted for Change in CMD.

^eAlso adjusted for Change in AUDIT.

*p ≤ 05; **p ≤ .01; ***p ≤ .001.

Sensitivity analysis

Given the nature of the original data, we could not confirm if participants were referring to the same relationship at both time points. To address this, we conducted a sensitivity analysis to investigate potential variations due to relationship type, limiting analyses to service personnel who reported being married at both time points (see supplementary materials, Tables 1–4).

Results

Sample characteristics

Among participants, 90.6% were men and 8.4% were women, with a mean (SD) age of 38.21 years (8.99 years) at p2 and 41.78 years (9.22) at p3 (see Table 1). Most participants were married with dependent children under 18 years of age at both phases and primarily regulars in the Army of NCO rank. At p2, 66.6% of personnel were serving, reducing to 35.6% at p3 as participants transitioned out of Service.

Compared to p2, participants in p3 were more likely to be married (86.4% vs. 74.5%), less likely to be cohabiting (14.5% vs.10.4%), and had higher education (39.6% with degree or above vs. 24.1%). A greater percentage of participants in p3 reported deploying to Iraq or Afghanistan (60.4% compared to 53.3%). Less than 20% at either phase met criteria for probable PTSD, CMD, or alcohol misuse at either phase, with an increase in probable PTSD slightly at p3 (4.9% vs. 2.9%).

Change in relationship satisfaction over time

Most participants 82.0% (n  = 3,137) reported no change in relationship satisfaction (Table 2), with 92.2% (n = 3,030) indicating they were satisfied/extremely satisfied. 8.0% (n  = 315) reported a positive change in their relationship satisfaction, and 10.0% (n  = 393) had a negative change. Relationship satisfaction over time was moderately positively correlated – participants who reported being extremely satisfied/satisfied with their relationship at p2 tended to report being extremely satisfied/satisfied at p3 (r(3843) = 0.27, p < .001).

Associations between change in relationship satisfaction and socio-demographic factors

After adjusting for socio-demographic factors significantly associated with changes in relationship satisfaction, respondents aged 40 years or more were less likely to report a negative change in relationship satisfaction compared to younger service personnel (Table 2; adj. OR 40–49 years = 0.75, 95% [0.56, 0.99], p = .04; adj. OR ≥50 years = 0.54, 95% CI [0.34–0.84], p < .01). Service personnel over 50 years were also less likely to report a positive change in relationship satisfaction compared to younger service personnel (adj. OR = 0.61, 95% CI [0.37–1.01], p = .05). Service personnel who reported two or more experiences of childhood family relationship adversity were more likely to endorse both positive (adj. OR = 2.53, 95% CI [1.76–3.65], p < .001) and negative (adj. OR = 1.60, 95% CI [1.18–2.16], p < .001) changes in relationship satisfaction compared to those who did not. There were no significant differences by gender or educational attainment (all p > .05).

Associations between military factors and changes in relationship satisfaction

After adjusting for socio-demographic and military factors significant in univariable models, Royal Navy/Royal Marines personnel were less likely to report a negative change in relationship satisfaction compared to Army personnel (adj. OR = 0.67, 95% CI [0.45, 0.98], p = .04) (Table 3). Service personnel who had left Service at p2 were less likely to endorse a negative change in relationship satisfaction compared to those still serving at p3 (adj. OR = 0.69, 95% CI [0.48–0.99], p = .05). No differences were found according to rank, regular/reserve status, deployment status, or combat role in the parent unit (all p > .05).

Associations between family circumstances and relationship satisfaction

Compared to married service personnel, participants in long-term relationships at p2 were significantly more likely to report positive changes in relationship satisfaction after adjusting for socio-demographic and military factors (adj. OR = 1.78, 95% CI [1.16–2.72], p < .01) (Table 4). Service personnel with children under 18 years at either phase were significantly more likely to report negative changes in relationship satisfaction compared to those with no children (adj. OR = 1.53, 95% CI [1.09–2.14], p = .01).

Associations between changes in mental health status, alcohol use, and relationship satisfaction

Compared to service personnel who did not meet criteria for probable PTSD or CMD at either phase, those who were persistent cases were significantly more likely to report both positive and negative changes in relationship satisfaction after adjusting for significant sociodemographic, military and family factors (Table 4; positive change (probable PTSD (adj. OR = 4.40, 95% CI [1.70–11.35], p < .01), probable CMD (adj. OR = 4.29, 95% CI [2.76–6.67], p < .001)); negative change (probable PTSD (adj. OR = 4.43, 95% CI [1.80–10.94], p = .001), probable CMD (adj. OR = 4.23, 95% CI [2.79–6.43], p < .001)). Service personnel endorsing remission of probable PTSD and CMD between p2 and p3 were also significantly more likely to report both positive and negative changes in relationship satisfaction compared to respondents who were not probable cases (positive change (probable PTSD (adj. OR = 10.95, 95% CI [5.19–23.12], p < .001), probable CMD (adj. OR = 3.82, 95% CI [2.65–5.52], p < .001); negative change (probable PTSD (adj. OR = 4.71, 95% CI [1.83–12.09], p = .001), probable CMD (adj. OR = 2.00, 95% CI [1.29–3.40], p < .01)). Service personnel who experienced new-onset probable PTSD or CMD were significantly more likely to report a negative change (probable PTSD (adj. OR = 4.57, 95% CI [2.75–7.60], p < .001), probable CMD (adj. OR = 3.34, 95% CI [2.37–4.69], p < .001)) in relationship satisfaction compared to those who were not probable cases at either phase.

Compared to service personnel who were not cases for alcohol misuse at either phase (AUDIT ≥16), those who were persistent cases at both p2 and p3 were more likely to report a positive change in relationship satisfaction (adj. OR = 2.65, 95% CI [1.42–4.95], p < .01), as were those who reported remission of alcohol misuse at p3 (adj. OR = 2.04, 95% CI [1.20, 3.47], p = .01). Service personnel who reported a new onset of alcohol misuse at p3 were more likely to report negative changes in relationship satisfaction (adj. OR = 2.89, 95% CI [1.79–4.47], p < .001) compared to service personnel who did not meet the criteria for alcohol misuse.

Sensitivity analysis

Results from the sensitivity analysis were broadly consistent with the main findings (see supplementary materials). Alcohol misuse remission was no longer significantly associated with negative changes in relationship satisfaction (p > .05); however, the effect size was similar to that in the main analysis, suggesting this may be due to a lack of power.

Discussion

These findings suggest that the majority of serving and ex-serving UK personnel experience stable relationship satisfaction over time, with 82% reporting no change between phase 2 and 3. This indicates that, for most service personnel, occupational demands, military career, deployments, and combat experience do not significantly affect perceptions of their relationships. Only a small proportion (10%) reported negative changes in relationship satisfaction, which were associated with stressors such as new-onset probable PTSD, CMD, alcohol misuse, and having dependent children. Conversely, positive changes, reported by 8% of the sample, were associated with alcohol misuse remission and persistent probable alcohol misuse.

Notably, several factors, including childhood family relationship adversity, persistent and remitting PTSD or CMD, and being of younger age (<40 years old), were associated with both positive and negative changes, highlighting the complexity of understanding and appreciating influences on relationship satisfaction among military populations. While the findings need to be replicated within multi-factorial measures of relationship satisfaction, these results challenge the idea that mental health is always a burden for military couples.

The role of socio-demographic, military characteristics, and family circumstances

Replicating prior research, age emerged as a predictor for changes in relationship satisfaction (Karney & Bradbury, 1997; Wilson & Stuchbury, 2010). Service personnel over 50 years of age were less likely to report a positive change than younger service personnel. possibly due to longer relationship length which has been linked to decreased relationship satisfaction over time (Burrell et al., 2006). However, older service personnel (≥40 years) were less likely to report negative changes over time compared to younger service personnel, possibly indicating more stable relationships among this group (Giordano et al., 2012). Parental status was associated with negative changes, highlighting the challenges dependent children can have on intimate relationships (Caska-Wallace et al., 2019; Keeling et al., 2015). This aligns with prior research conducted among both general and military populations (Caska-Wallace et al., 2019; Keeling et al., 2015; Padilla-Walker et al., 2014).

Service personnel in long-term relationships were more likely to report a positive change compared to their married counterparts, a finding that is in contrast with prior cohort and general population findings (Keeling et al., 2015; Wilson & Stuchbury, 2010). This may reflect new relationships formed between the phases and warrants further investigation. Childhood family relationship adversity was associated with positive and negative changes among service personnel, contradicting trends found among the general population (Whisman, 2006). Specifically, previous findings among the general population have shown that experiencing such adversity is associated with lower intimacy and emotional difficulties during adulthood (DiLillo et al., 2009; Whisman, 2006). This difference between populations highlights the need for additional research exploring how military-specific factors may moderate the effects of childhood adversity on relationship satisfaction.

Consistent with previous cohort phases, no significant gender differences in relationship satisfaction were found (Keeling et al., 2015), though future research should continue focusing on women’s relationship experiences in Service (Southwell & macdermid Wadsworth, 2016; Vogt et al., 2011), as our study was not adequately powered to detect differences across gender. Furthermore, educational attainment did not predict changes in relationship satisfaction, in contrast to findings in the general population (Wilson & Stuchbury, 2010) and US military populations (Riviere et al., 2012), indicating that it might not be a strong predictor among UK service personnel.

Few military-specific factors were associated with changes in relationship satisfaction. Service and ex-service personnel in the Royal Navy and Royal Marines were less likely to report a negative change compared to Army personnel, potentially reflecting differences in occupational demands such as mobility and satisfaction with Service life (Ministry of Defence, 2020). Consistent with previous findings among this cohort (Keeling et al., 2015), military rank, deployment, and combat were not associated with changes in relationship satisfaction. The lack of differences between combat and deployment may be attributed to the time elapsed since these events, while the absence of differences across military ranks contrasts with US literature that suggests that lower ranks often report greater relationship dissatisfaction (Anderson et al., 2011). This may be due to other factors such as social class or income being better predictors of relationship outcomes in this dataset. Furthermore, no differences in relationship satisfaction were found between serving and ex-serving personnel, possibly due to data collection not fully capturing relationship challenges during transition.

Associations with changes in mental health and alcohol misuse over time

Consistent with prior literature (Allen et al., 2010; Campbell & Renshaw, 2013; Knobloch & Theiss, 2011), persistent probable PTSD and CMD cases, as well as the onset of probable PTSD, CMD, and alcohol misuse, were associated with an increased likelihood of reporting negative changes in relationship satisfaction. These findings potentially reflect the disruptive impact of such symptoms on relationship dynamics, such as emotional numbing, emotional withdrawal, hyperarousal, and deterioration of communication quality (Foran et al., 2013; Miller et al., 2013).

However, our findings also indicate that some service personnel with persistent probable PTSD and CMD symptoms also reported a positive change. These findings may be attributed to the development of better-coping mechanisms, greater emotional resilience, or the presence of a supportive partner or community resources through challenging times may have enhanced service personnel’s appreciation of their relationship face (Allen et al., 2021; Thandi et al., 2017; Vogt et al., 2011). In contrast to prior studies (Mattiko et al., 2011; Meis et al., 2010; Riviere et al., 2012), persistent alcohol misuse was associated with positive changes for some, possibly reflecting complex coping dynamics. Remission from poor mental health and alcohol misuse was also predicted to increase the likelihood of reporting positive changes, which could be due to improved emotional availability and communication, and addressing substance use issues (Edwards-Stewart et al., 2018; Leonard & Eiden, 2007). PTSD and CMD remission were also associated with reporting negative changes, possibly due to the re-adjusting period from a caregiver-patient dynamic back to an intimate relationship (Thandi et al., 2017).

Strengths and limitations

This study contributes to the existing literature by investigating both positive and negative changes in relationship satisfaction across two time points among UK service personnel. It provides valuable insight into the nuances of relationships within the military community and how certain factors that may be stressors for some couples may be opportunities for others. The findings highlight the importance of investigating both challenges and benefits associated with military service, rather than solely focusing on the adverse effects and appreciating that one size does not fit all military couples or families.

The findings should be considered in light of the study’s limitations. While the findings provide insights into relationship satisfaction over time for military populations, especially in the UK, the assessment of relationship satisfaction was limited to a single-item measure that was subsequently categorized. The use of a one-item measure may mean that other aspects of relationship satisfaction were not captured and that other factors known to contribute to multi-factorial understandings of intimate relationships, such as commitment, investment in the relationship, self-disclosure, and emotional intimacy were missed (Karney & Crown, 2007). Categorization facilitated analyses across time and allowed comparison with prior UK studies (Keeling et al., 2015) but is likely to have obscured more subtle patterns of change – for instance, participants experiencing an increase in positive assessment of their relationship (e.g., satisfied to extremely satisfied) were grouped with those potentially experiencing changes (e.g., dissatisfied to extremely satisfied). However, we note a great deal of stability within these categories, with 90% of respondents within the “no change” category indicating that they were satisfied or extremely satisfied with their relationship at both time points. Some degree of stability is also demonstrated in the positive (>70% moving from dissatisfied/extremely dissatisfied to satisfied/extremely satisfied) and negative (>60% moving from satisfied/extremely satisfied to dissatisfied/extremely dissatisfied) change groups. We recommend that future studies explore relationship dynamics among UK service personnel and ex-serving personnel using multi-dimensional measures to provide a more comprehensive understanding of relationship satisfaction and its predictors over time.

Due to the nature of the data, it was not possible to determine whether participants were reporting on the same relationship at both time points, although sensitivity analyses suggest findings hold for potentially more stable married relationships. While findings relating to sociodemographic factors were broadly in line with prior research, military factors were somewhat more mixed, possibly due to the time since personnel experienced (combat) deployments and the nature of UK service compared to that of the US. These findings imply that although rank could impact certain influences on relationships, such as stability and financial security, it may not directly impact on relationship satisfaction itself. Transition stressors or improvements in relationship satisfaction after leaving Service were also not identified. Ex-serving personnel who participated may have been in relationships that successfully managed the challenges of transitioning from military to civilian life or left relationships that were not able to manage this change.

The use of different PTSD measures across timepoints introduces additional considerations. Overall PTSD caseness over time was determined using participant caseness on the PCL-C (≥50) at p2 and the PCL-5 (≥38) at p3. While prior research has demonstrated comparability between these measures and their respective cutoffs (LeardMann et al., 2021; Moshier et al., 2019), it is possible that differences in the structure and scoring of these measures affected the consistency of this comparison, although this is likely to have minimal effect on overall findings.

Lastly, while this study provides valuable insights, its generalizability is limited by the age of the dataset and the reliance on self-report measures, which only explored the perspective of the service personnel. Changes in military culture and policies since data collection, such as opening military housing to non-married couples, may have influenced relationship dynamics and satisfaction differently. Future studies should aim to replicate these findings using contemporary data and include the perspective from partners of service personnel.

Implications

Despite the study’s limitations, our findings suggest that most service personnel reported stable relationship satisfaction over time. Such couples may be able to provide insights for others into how to manage the daily stressors of military life and could form the basis for a short information pack for new couples joining the Armed Forces community. However, given that mental health outcomes significantly predicted changes in relationship satisfaction, both positive and negative, it may be useful to provide information and training for both married and unmarried couples. Such initiatives should emphasize communication and understanding of the difficulties and challenges each respective partner may face, as well as the potential for growth and acceptance, especially for the newly diagnosed (Gottman et al., 2011; Saltzman et al., 2011). An adapted version of the Families OverComing Under Stress (FOCUS, Nathanson Family Resilience Center) training program may be proposed as a potential family intervention or further roll-out of the Royal Air Force-led Building Stronger Families post-evaluation.

Conclusion

Overall, this research highlights the complexity of relationships among service personnel and underscores the importance of understanding the roles of various socio-demographic, military, and family factors in relationship satisfaction. While most service personnel report stable relationship satisfaction, factors such as age, childhood family adversity, parental status, and mental health changes can influence relationship satisfaction, positively or negatively, over time. Crucially, some of the challenges faced by service personnel may lead to positive changes in relationship satisfaction for some couples, suggesting adverse circumstances need not always have a negative impact on relationships and could, in fact, promote growth and development. These findings have practical implications for the design and implementation of interventions aimed at improving relationship outcomes for service personnel and their families and encouraging retention.

Funding Statement

This project was funded by the 2020 King’s Undergraduate Research Fellowship program, King’s College London.

Disclosure statement

NTF is a trustee (unpaid) for Help for Heroes, a nonprofit supporting UK Armed Forces and is a member of the Independent Scientific Pandemic Insights Group on Behaviours (SPI-B). Both NTF and HB are paid in part by a grant from the UK Ministry of Defence.

Data availability statement

The data for this study may be available on request, subject to approvals from KCMHR senior leadership. The data are not publicly available due to security restrictions.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/08995605.2025.2527440.